2021-05-05 14:43:56 +02:00
{
"cells": [
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Uczenie maszynowe\n",
"# 4. Sieci neuronowe –
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [],
"source": [
"# Przydatne importy\n",
"\n",
"import matplotlib\n",
"import matplotlib.pyplot as plt\n",
"import numpy as np\n",
"\n",
"%matplotlib inline"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## 4.1. Perceptron"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"https://www.youtube.com/watch?v=cNxadbrN_aI"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img style=\"margin: auto\" heighth=\"100%\" src=\"http://m.natemat.pl/b94a41cd7322e1b8793e4644e5f82683,641,0,0,0.png\" alt=\"Frank Rosenblatt\"/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img style=\"margin: auto\" src=\"http://m.natemat.pl/02943a7dc0f638d786b78cd5c9e75742,641,0,0,0.png\" heighth=\"100%\" alt=\"Frank Rosenblatt\"/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img style=\"margin: auto\" heighth=\"100%\" src=\"https://upload.wikimedia.org/wikipedia/en/5/52/Mark_I_perceptron.jpeg\" alt=\"perceptron\"/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Pierwszy perceptron liniowy\n",
"\n",
"* Frank Rosenblatt, 1957\n",
"* aparat fotograficzny podłączony do 400 fotokomórek (rozdzielczość obrazu: 20 x 20)\n",
"* wagi –
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Uczenie perceptronu\n",
"\n",
"Cykl uczenia perceptronu Rosenblatta:\n",
"\n",
"1. Sfotografuj planszę z kolejnym obiektem.\n",
"1. Zaobserwuj, która lampka zapaliła się na wyjściu.\n",
"1. Sprawdź, czy to jest właściwa lampka.\n",
"1. Wyślij sygnał „nagrody” lub „kary”."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Funkcja aktywacji\n",
"\n",
"Funkcja bipolarna:\n",
"\n",
"$$ g(z) = \\left\\{ \n",
"\\begin{array}{rl}\n",
"1 & \\textrm{gdy $z > \\theta_0$} \\\\\n",
"-1 & \\textrm{wpp.}\n",
"\\end{array}\n",
"\\right. $$\n",
"\n",
"gdzie $z = \\theta_0x_0 + \\ldots + \\theta_nx_n$,<br/>\n",
"$\\theta_0$ to próg aktywacji,<br/>\n",
"$x_0 = 1$. "
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [],
"source": [
"def bipolar_plot():\n",
" matplotlib.rcParams.update({'font.size': 16})\n",
"\n",
" plt.figure(figsize=(8,5))\n",
" x = [-1,-.23,1] \n",
" y = [-1, -1, 1]\n",
" plt.ylim(-1.2,1.2)\n",
" plt.xlim(-1.2,1.2)\n",
" plt.plot([-2,2],[1,1], color='black', ls=\"dashed\")\n",
" plt.plot([-2,2],[-1,-1], color='black', ls=\"dashed\")\n",
" plt.step(x, y, lw=3)\n",
" ax = plt.gca()\n",
" ax.spines['right'].set_color('none')\n",
" ax.spines['top'].set_color('none')\n",
" ax.xaxis.set_ticks_position('bottom')\n",
" ax.spines['bottom'].set_position(('data',0))\n",
" ax.yaxis.set_ticks_position('left')\n",
" ax.spines['left'].set_position(('data',0))\n",
"\n",
" plt.annotate(r'$\\theta_0$',\n",
" xy=(-.23,0), xycoords='data',\n",
" xytext=(-50, +50), textcoords='offset points', fontsize=26,\n",
" arrowprops=dict(arrowstyle=\"->\"))\n",
"\n",
" plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"data": {
"image/png": "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
"text/plain": [
"<Figure size 576x360 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"bipolar_plot()"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Perceptron –
"\n",
"<img src=\"perceptron.png\" width=\"60%\"/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### Perceptron –
"\n",
"1. Ustal wartości początkowe $\\theta$ (wektor 0 lub liczby losowe blisko 0).\n",
"1. Dla każdego przykładu $(x^{(i)}, y^{(i)})$, dla $i=1,\\ldots,m$\n",
" * Oblicz wartość wyjścia $o^{(i)} = g(\\theta^{T}x^{(i)}) = g(\\sum_{j=0}^{n} \\theta_jx_j^{(i)})$\n",
" * Wykonaj aktualizację
" $$ \\theta := \\theta + \\Delta \\theta $$\n",
" $$ \\Delta \\theta = \\alpha(y^{(i)}-o^{(i)})x^{(i)} $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"$$\\theta_j := \\theta_j + \\Delta \\theta_j $$\n",
"\n",
"Jeżeli przykład został sklasyfikowany **poprawnie**:\n",
"\n",
"* $y^{(i)}=1$ oraz $o^{(i)}=1$ : $$\\Delta\\theta_j = \\alpha(1 - 1)x_j^{(i)} = 0$$\n",
"* $y^{(i)}=-1$ oraz $o^{(i)}=-1$ : $$\\Delta\\theta_j = \\alpha(-1 - -1)x_j^{(i)} = 0$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"Czyli: jeżeli trafiłeś, to nic nie zmieniaj."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"$$\\theta_j := \\theta_j + \\Delta \\theta_j $$\n",
"\n",
"Jeżeli przykład został sklasyfikowany **niepoprawnie**:\n",
"\n",
"* $y^{(i)}=1$ oraz $o^{(i)}=-1$ : $$\\Delta\\theta_j = \\alpha(1 - -1)x_j^{(i)} = 2 \\alpha x_j^{(i)}$$\n",
"* $y^{(i)}=-1$ oraz $o^{(i)}=1$ : $$\\Delta\\theta_j = \\alpha(-1 - 1)x_j^{(i)} = -2 \\alpha x_j^{(i)}$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"Czyli: przesuń wagi w odpowiednią stronę."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Perceptron –
"\n",
"* intuicyjny i prosty\n",
"* łatwy w implementacji\n",
"* jeżeli dane można liniowo oddzielić, algorytm jest zbieżny w skończonym czasie"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Perceptron –
"\n",
"* jeżeli danych nie można oddzielić liniowo, algorytm nie jest zbieżny"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [],
"source": [
"def plot_perceptron():\n",
" plt.figure(figsize=(12,3))\n",
"\n",
" plt.subplot(131)\n",
" plt.ylim(-0.2,1.2)\n",
" plt.xlim(-0.2,1.2)\n",
"\n",
" plt.title('AND')\n",
" plt.plot([1,0,0], [0,1,0], 'ro', markersize=10)\n",
" plt.plot([1], [1], 'go', markersize=10)\n",
"\n",
" ax = plt.gca()\n",
" ax.spines['right'].set_color('none')\n",
" ax.spines['top'].set_color('none')\n",
" ax.xaxis.set_ticks_position('none')\n",
" ax.spines['bottom'].set_position(('data',0))\n",
" ax.yaxis.set_ticks_position('none')\n",
" ax.spines['left'].set_position(('data',0))\n",
"\n",
" plt.xticks(np.arange(0, 2, 1.0))\n",
" plt.yticks(np.arange(0, 2, 1.0))\n",
"\n",
"\n",
" plt.subplot(132)\n",
" plt.ylim(-0.2,1.2)\n",
" plt.xlim(-0.2,1.2)\n",
"\n",
" plt.plot([1,0,1], [0,1,1], 'go', markersize=10)\n",
" plt.plot([0], [0], 'ro', markersize=10)\n",
"\n",
" ax = plt.gca()\n",
" ax.spines['right'].set_color('none')\n",
" ax.spines['top'].set_color('none')\n",
" ax.xaxis.set_ticks_position('none')\n",
" ax.spines['bottom'].set_position(('data',0))\n",
" ax.yaxis.set_ticks_position('none')\n",
" ax.spines['left'].set_position(('data',0))\n",
"\n",
" plt.title('OR')\n",
" plt.xticks(np.arange(0, 2, 1.0))\n",
" plt.yticks(np.arange(0, 2, 1.0))\n",
"\n",
"\n",
" plt.subplot(133)\n",
" plt.ylim(-0.2,1.2)\n",
" plt.xlim(-0.2,1.2)\n",
"\n",
" plt.title('XOR')\n",
" plt.plot([1,0], [0,1], 'go', markersize=10)\n",
" plt.plot([0,1], [0,1], 'ro', markersize=10)\n",
"\n",
" ax = plt.gca()\n",
" ax.spines['right'].set_color('none')\n",
" ax.spines['top'].set_color('none')\n",
" ax.xaxis.set_ticks_position('none')\n",
" ax.spines['bottom'].set_position(('data',0))\n",
" ax.yaxis.set_ticks_position('none')\n",
" ax.spines['left'].set_position(('data',0))\n",
"\n",
" plt.xticks(np.arange(0, 2, 1.0))\n",
" plt.yticks(np.arange(0, 2, 1.0))\n",
"\n",
" plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"data": {
"image/png": "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
"text/plain": [
"<Figure size 864x216 with 3 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"plot_perceptron()"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Funkcje aktywacji\n",
"\n",
"Zamiast funkcji bipolarnej możemy zastosować funkcję sigmoidalną jako funkcję aktywacji."
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [],
"source": [
"def plot_activation_functions():\n",
" plt.figure(figsize=(16,7))\n",
" plt.subplot(121)\n",
" x = [-2,-.23,2] \n",
" y = [-1, -1, 1]\n",
" plt.ylim(-1.2,1.2)\n",
" plt.xlim(-2.2,2.2)\n",
" plt.plot([-2,2],[1,1], color='black', ls=\"dashed\")\n",
" plt.plot([-2,2],[-1,-1], color='black', ls=\"dashed\")\n",
" plt.step(x, y, lw=3)\n",
" ax = plt.gca()\n",
" ax.spines['right'].set_color('none')\n",
" ax.spines['top'].set_color('none')\n",
" ax.xaxis.set_ticks_position('bottom')\n",
" ax.spines['bottom'].set_position(('data',0))\n",
" ax.yaxis.set_ticks_position('left')\n",
" ax.spines['left'].set_position(('data',0))\n",
"\n",
" plt.annotate(r'$\\theta_0$',\n",
" xy=(-.23,0), xycoords='data',\n",
" xytext=(-50, +50), textcoords='offset points', fontsize=26,\n",
" arrowprops=dict(arrowstyle=\"->\"))\n",
"\n",
" plt.subplot(122)\n",
" x2 = np.linspace(-2,2,100)\n",
" y2 = np.tanh(x2+ 0.23)\n",
" plt.ylim(-1.2,1.2)\n",
" plt.xlim(-2.2,2.2)\n",
" plt.plot([-2,2],[1,1], color='black', ls=\"dashed\")\n",
" plt.plot([-2,2],[-1,-1], color='black', ls=\"dashed\")\n",
" plt.plot(x2, y2, lw=3)\n",
" ax = plt.gca()\n",
" ax.spines['right'].set_color('none')\n",
" ax.spines['top'].set_color('none')\n",
" ax.xaxis.set_ticks_position('bottom')\n",
" ax.spines['bottom'].set_position(('data',0))\n",
" ax.yaxis.set_ticks_position('left')\n",
" ax.spines['left'].set_position(('data',0))\n",
"\n",
" plt.annotate(r'$\\theta_0$',\n",
" xy=(-.23,0), xycoords='data',\n",
" xytext=(-50, +50), textcoords='offset points', fontsize=26,\n",
" arrowprops=dict(arrowstyle=\"->\"))\n",
"\n",
" plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"data": {
"image/png": "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
"text/plain": [
"<Figure size 1152x504 with 2 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"plot_activation_functions()"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Perceptron a regresja liniowa"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img src=\"reglin.png\" width=\"70%\"/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Uczenie regresji liniowej:\n",
"* Model: $$h_{\\theta}(x) = \\sum_{i=0}^n \\theta_ix_i$$\n",
"* Funkcja kosztu (błąd średniokwadratowy): $$J(\\theta) = \\frac{1}{m} \\sum_{i=1}^{m} (h_{\\theta}(x^{(i)}) - y^{(i)})^2$$\n",
"* Po obliczeniu $\\nabla J(\\theta)$ - zwykły SGD."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Perceptron a dwuklasowa regresja logistyczna"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<img src=\"reglog.png\" width=\"60%\"/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Uczenie dwuklasowej regresji logistycznej:\n",
"* Model: $h_{\\theta}(x) = \\sigma(\\sum_{i=0}^n \\theta_ix_i) = P(1|x,\\theta)$\n",
"* Funkcja kosztu (entropia krzyżowa): $$\\begin{eqnarray} J(\\theta) &=& -\\frac{1}{m} \\sum_{i=1}^{m} \\big( y^{(i)}\\log P(1|x^{(i)},\\theta) \\\\ && + (1-y^{(i)})\\log(1-P(1|x^{(i)},\\theta)) \\big) \\end{eqnarray}$$\n",
"* Po obliczeniu $\\nabla J(\\theta)$ - zwykły SGD."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Perceptron a wieloklasowa regresja logistyczna"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img src=\"multireglog.png\" width=\"40%\"/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Wieloklasowa regresja logistyczna\n",
"* Model (dla $c$ klasyfikatorów binarnych): \n",
"$$\\begin{eqnarray}\n",
"h_{(\\theta^{(1)},\\dots,\\theta^{(c)})}(x) &=& \\mathrm{softmax}(\\sum_{i=0}^n \\theta_{i}^{(1)}x_i, \\ldots, \\sum_{i=0}^n \\theta_i^{(c)}x_i) \\\\ \n",
"&=& \\left[ P(k|x,\\theta^{(1)},\\dots,\\theta^{(c)}) \\right]_{k=1,\\dots,c} \n",
"\\end{eqnarray}$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Funkcja kosztu (**przymując model regresji binarnej**): $$\\begin{eqnarray} J(\\theta^{(k)}) &=& -\\frac{1}{m} \\sum_{i=1}^{m} \\big( y^{(i)}\\log P(k|x^{(i)},\\theta^{(k)}) \\\\ && + (1-y^{(i)})\\log P(\\neg k|x^{(i)},\\theta^{(k)}) \\big) \\end{eqnarray}$$\n",
"* Po obliczeniu $\\nabla J(\\theta)$, **c-krotne** uruchomienie SGD, zastosowanie $\\mathrm{softmax}(X)$ do niezależnie uzyskanych klasyfikatorów binarnych."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Przyjmijmy: \n",
"$$ \\Theta = (\\theta^{(1)},\\dots,\\theta^{(c)}) $$\n",
"\n",
"$$h_{\\Theta}(x) = \\left[ P(k|x,\\Theta) \\right]_{k=1,\\dots,c}$$\n",
"\n",
"$$\\delta(x,y) = \\left\\{\\begin{array}{cl} 1 & \\textrm{gdy } x=y \\\\ 0 & \\textrm{wpp.}\\end{array}\\right.$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Wieloklasowa funkcja kosztu $J(\\Theta)$ (kategorialna entropia krzyżowa):\n",
"$$ J(\\Theta) = -\\frac{1}{m}\\sum_{i=1}^{m}\\sum_{k=1}^{c} \\delta({y^{(i)},k}) \\log P(k|x^{(i)},\\Theta) $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Gradient $\\nabla J(\\Theta)$:\n",
"$$ \\dfrac{\\partial J(\\Theta)}{\\partial \\Theta_{j,k}} = -\\frac{1}{m}\\sum_{i = 1}^{m} (\\delta({y^{(i)},k}) - P(k|x^{(i)}, \\Theta)) x^{(i)}_j \n",
"$$\n",
"\n",
"* Liczymy wszystkie wagi jednym uruchomieniem SGD"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"## Podsumowanie\n",
"\n",
"* W przypadku jednowarstowej sieci neuronowej wystarczy znać gradient funkcji kosztu.\n",
"* Wtedy liczymy tak samo jak w przypadku regresji liniowej, logistycznej, wieloklasowej logistycznej itp. (wymienione modele to szczególne przypadki jednowarstwowych sieci neuronowych).\n",
"* Regresja liniowa i binarna regresja logistyczna to jeden neuron.\n",
"* Wieloklasowa regresja logistyczna to tyle neuronów, ile klas."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"Funkcja aktywacji i funkcja kosztu są **dobierane do problemu**."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## 4.2. Funkcje aktywacji"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
2022-05-21 08:03:16 +02:00
"\n",
2021-05-05 14:43:56 +02:00
"* Każda funkcja aktywacji ma swoje zalety i wady.\n",
"* Różne rodzaje funkcji aktywacji nadają się do różnych zastosowań."
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/pawel/.local/lib/python2.7/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
" from ._conv import register_converters as _register_converters\n",
"Using TensorFlow backend.\n"
]
}
],
"source": [
"%matplotlib inline\n",
"\n",
"import math\n",
"import matplotlib.pyplot as plt\n",
"import numpy as np\n",
"import random\n",
"\n",
"import keras\n",
"from keras.datasets import mnist\n",
"from keras.models import Sequential\n",
"from keras.layers import Dense, Dropout, SimpleRNN, LSTM\n",
"from keras.optimizers import Adagrad, Adam, RMSprop, SGD\n",
"\n",
"from IPython.display import YouTubeVideo"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [],
"source": [
"def plot(fun):\n",
" x = np.arange(-3.0, 3.0, 0.01)\n",
" y = [fun(x_i) for x_i in x]\n",
" fig = plt.figure(figsize=(14, 7))\n",
" ax = fig.add_subplot(111)\n",
" fig.subplots_adjust(left=0.1, right=0.9, bottom=0.1, top=0.9)\n",
" ax.set_xlim(-3.0, 3.0)\n",
" ax.set_ylim(-1.5, 1.5)\n",
" ax.grid()\n",
" ax.plot(x, y)\n",
" plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Funkcja logistyczna"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"$$ g(x) = \\frac{1}{1 + e^{-x}} $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* Przyjmuje wartości z przedziału $(0, 1)$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### Funkcja logistyczna –
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAA1cAAAG2CAYAAACTRXz+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nY6rx/OZUzk+/vbDEtr6
"text/plain": [
"<matplotlib.figure.Figure at 0x7fdda9490fd0>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot(lambda x: 1 / (1 + math.exp(-x)))"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Tangens hiperboliczny"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"$$ g(x) = \\tanh x = \\frac{e^{x} - e^{-x}}{e^{x} + e^{-x}} $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* Przyjmuje wartości z przedziału $(-1, 1)$.\n",
"* Powstaje z funkcji logistycznej przez przeskalowanie i przesunięcie."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### Tangens hiperboliczny –
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAA1cAAAG2CAYAAACTRXz+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMQE4iLgCAABwgLVW6/Z
"text/plain": [
"<matplotlib.figure.Figure at 0x7fdda93e9590>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot(lambda x: math.tanh(x))"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### ReLU (*Rectifier Linear Unit*)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"$$ g(x) = \\max(0, x) $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### ReLU –
"* Mniej podatna na problem zanikającego gradientu (*vanishing gradient*) niż funkcje sigmoidalne, dzięki czemu SGD jest szybciej zbieżna.\n",
"* Prostsze obliczanie gradientu.\n",
"* Dzięki zerowaniu ujemnych wartości, wygasza neurony, „rozrzedzając” sieć (*sparsity*), co przyspiesza obliczenia."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"#### ReLU –
"* Dla dużych wartości gradient może „eksplodować”.\n",
"* „Wygaszanie” neuronów."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### ReLU –
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAA1cAAAG2CAYAAACTRXz+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCoCVJawA6ElcAbCShBU
"text/plain": [
"<matplotlib.figure.Figure at 0x7fdda936c6d0>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot(lambda x: max(0, x))"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Softplus"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"$$ g(x) = \\log(1 + e^{x}) $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* Wygładzona wersja ReLU."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### Softplus –
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAA1cAAAG2CAYAAACTRXz+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nG9bGOxpaO3XH0k26b9l
"text/plain": [
"<matplotlib.figure.Figure at 0x7fdde8452e10>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot(lambda x: math.log(1 + math.exp(x)))"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Problem zanikającego gradientu (*vanishing gradient problem*)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* Sigmoidalne funkcje aktywacji ograniczają wartości na wyjściach neuronów do niewielkich przedziałów ($(-1, 1)$, $(0, 1)$ itp.).\n",
"* Jeżeli sieć ma wiele warstw, to podczas propagacji wstecznej mnożymy przez siebie wiele małych wartości → obliczony gradient jest mały.\n",
"* Im więcej warstw, tym silniejszy efekt zanikania."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### Sposoby na zanikający gradient\n",
"\n",
"* Modyfikacja algorytmu optymalizacji (*RProp*, *RMSProp*)\n",
"* Użycie innej funckji aktywacji (ReLU, softplus)\n",
"* Dodanie warstw *dropout*\n",
"* Nowe architektury (LSTM itp.)\n",
"* Więcej danych, zwiększenie mocy obliczeniowej"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## 4.3. Wielowarstwowe sieci neuronowe\n",
"\n",
"czyli _Artificial Neural Networks_ (ANN) lub _Multi-Layer Perceptrons_ (MLP)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img src=\"nn1.png\" width=\"70%\"/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Architektura sieci\n",
"\n",
"* Sieć neuronowa jako graf neuronów. \n",
"* Organizacja sieci przez warstwy.\n",
"* Najczęściej stosowane są sieci jednokierunkowe i gęste."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* $n$-warstwowa sieć neuronowa ma $n+1$ warstw (nie liczymy wejścia).\n",
"* Rozmiary sieci określane poprzez liczbę neuronów lub parametrów."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Sieć neuronowa jednokierunkowa (*feedforward*)\n",
"\n",
"* Mając daną $n$-warstwową sieć neuronową oraz jej parametry $\\Theta^{(1)}, \\ldots, \\Theta^{(L)} $ oraz $\\beta^{(1)}, \\ldots, \\beta^{(L)} $ liczymy:<br/><br/> \n",
"$$a^{(l)} = g^{(l)}\\left( a^{(l-1)} \\Theta^{(l)} + \\beta^{(l)} \\right). $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img src=\"nn2.png\" width=70%/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Funkcje $g^{(l)}$ to tzw. **funkcje aktywacji**.<br/>\n",
"Dla $i = 0$ przyjmujemy $a^{(0)} = \\mathrm{x}$ (wektor wierszowy cech) oraz $g^{(0)}(x) = x$ (identyczność)."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* Parametry $\\Theta$ to wagi na połączeniach miedzy neuronami dwóch warstw.<br/>\n",
"Rozmiar macierzy $\\Theta^{(l)}$, czyli macierzy wag na połączeniach warstw $a^{(l-1)}$ i $a^{(l)}$, to $\\dim(a^{(l-1)}) \\times \\dim(a^{(l)})$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Parametry $\\beta$ zastępują tutaj dodawanie kolumny z jedynkami do macierzy cech.<br/>Macierz $\\beta^{(l)}$ ma rozmiar równy liczbie neuronów w odpowiedniej warstwie, czyli $1 \\times \\dim(a^{(l)})$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* **Klasyfikacja**: dla ostatniej warstwy $L$ (o rozmiarze równym liczbie klas) przyjmuje się $g^{(L)}(x) = \\mathop{\\mathrm{softmax}}(x)$.\n",
"* **Regresja**: pojedynczy neuron wyjściowy jak na obrazku. Funkcją
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Pozostałe funkcje aktywacji najcześciej mają postać sigmoidy, np. sigmoidalna, tangens hiperboliczny.\n",
"* Mogą mieć też inny kształt, np. ReLU, leaky ReLU, maxout."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
2021-05-05 15:23:24 +02:00
"### Jak uczyć sieci neuronowe?"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* W poznanych do tej pory algorytmach (regresja liniowa, regresja logistyczna) do uczenia używaliśmy funkcji kosztu, jej gradientu oraz algorytmu gradientu prostego (GD/SGD)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* Dla sieci neuronowych potrzebowalibyśmy również znaleźć gradient funkcji kosztu."
2021-05-05 14:43:56 +02:00
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
2021-05-05 15:23:24 +02:00
"* Sprowadza się to do bardziej ogólnego problemu:<br/>jak obliczyć gradient $\\nabla f(x)$ dla danej funkcji $f$ i wektora wejściowego $x$?"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## 4.4. Metoda propagacji wstecznej –
2021-05-05 14:43:56 +02:00
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
2021-05-05 15:23:24 +02:00
"### Pochodna funkcji\n",
2021-05-05 14:43:56 +02:00
"\n",
2021-05-05 15:23:24 +02:00
"* **Pochodna** mierzy, jak szybko zmienia się wartość funkcji względem zmiany jej argumentów:\n",
2021-05-05 14:43:56 +02:00
"\n",
2021-05-05 15:23:24 +02:00
"$$ \\frac{d f(x)}{d x} = \\lim_{h \\to 0} \\frac{ f(x + h) - f(x) }{ h } $$"
2021-05-05 14:43:56 +02:00
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
2021-05-05 15:23:24 +02:00
"### Pochodna cząstkowa i gradient\n",
2021-05-05 14:43:56 +02:00
"\n",
2021-05-05 15:23:24 +02:00
"* **Pochodna cząstkowa** mierzy, jak szybko zmienia się wartość funkcji względem zmiany jej *pojedynczego argumentu*."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* **Gradient** to wektor pochodnych cząstkowych:\n",
2021-05-05 14:43:56 +02:00
"\n",
2021-05-05 15:23:24 +02:00
"$$ \\nabla f = \\left( \\frac{\\partial f}{\\partial x_1}, \\ldots, \\frac{\\partial f}{\\partial x_n} \\right) $$"
2021-05-05 14:43:56 +02:00
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
2021-05-05 15:23:24 +02:00
"#### Gradient –
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"$$ f(x_1, x_2) = x_1 + x_2 \\qquad \\to \\qquad \\frac{\\partial f}{\\partial x_1} = 1, \\quad \\frac{\\partial f}{\\partial x_2} = 1, \\quad \\nabla f = (1, 1) $$ "
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"$$ f(x_1, x_2) = x_1 \\cdot x_2 \\qquad \\to \\qquad \\frac{\\partial f}{\\partial x_1} = x_2, \\quad \\frac{\\partial f}{\\partial x_2} = x_1, \\quad \\nabla f = (x_2, x_1) $$ "
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"$$ f(x_1, x_2) = \\max(x_1 + x_2) \\hskip{12em} \\\\\n",
"\\to \\qquad \\frac{\\partial f}{\\partial x_1} = \\mathbb{1}_{x \\geq y}, \\quad \\frac{\\partial f}{\\partial x_2} = \\mathbb{1}_{y \\geq x}, \\quad \\nabla f = (\\mathbb{1}_{x \\geq y}, \\mathbb{1}_{y \\geq x}) $$ "
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Własności pochodnych cząstkowych\n",
2021-05-05 14:43:56 +02:00
"\n",
2021-05-05 15:23:24 +02:00
"Jezeli $f(x, y, z) = (x + y) \\, z$ oraz $x + y = q$, to:\n",
"$$f = q z,\n",
"\\quad \\frac{\\partial f}{\\partial q} = z,\n",
"\\quad \\frac{\\partial f}{\\partial z} = q,\n",
"\\quad \\frac{\\partial q}{\\partial x} = 1,\n",
"\\quad \\frac{\\partial q}{\\partial y} = 1 $$"
2021-05-05 14:43:56 +02:00
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
2021-05-05 15:23:24 +02:00
"slide_type": "subslide"
2021-05-05 14:43:56 +02:00
}
},
"source": [
2021-05-05 15:23:24 +02:00
"### Reguła łańcuchowa\n",
"\n",
"$$ \\frac{\\partial f}{\\partial x} = \\frac{\\partial f}{\\partial q} \\, \\frac{\\partial q}{\\partial x},\n",
"\\quad \\frac{\\partial f}{\\partial y} = \\frac{\\partial f}{\\partial q} \\, \\frac{\\partial q}{\\partial y} $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Propagacja wsteczna –
2021-05-05 14:43:56 +02:00
]
},
{
"cell_type": "code",
2021-05-05 15:23:24 +02:00
"execution_count": 2,
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
2021-05-05 14:43:56 +02:00
"outputs": [],
2021-05-05 15:23:24 +02:00
"source": [
"# Dla ustalonego wejścia\n",
"x = -2; y = 5; z = -4"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"(3, -12)\n"
]
}
],
"source": [
"# Krok w przód\n",
"q = x + y\n",
"f = q * z\n",
"print(q, f)"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[-4, -4, 3]\n"
]
}
],
"source": [
"# Propagacja wsteczna dla f = q * z\n",
"dz = q\n",
"dq = z\n",
"# Propagacja wsteczna dla q = x + y\n",
"dx = 1 * dq # z reguły łańcuchowej\n",
"dy = 1 * dq # z reguły łańcuchowej\n",
"print([dx, dy, dz])"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img src=\"exp1.png\" />"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Właśnie tak wygląda obliczanie pochodnych metodą propagacji wstecznej!"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* Spróbujmy czegoś bardziej skomplikowanego:<br/>metodą propagacji wstecznej obliczmy pochodną funkcji sigmoidalnej."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Propagacja wsteczna –
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"Funkcja sigmoidalna:\n",
"\n",
"$$f(\\theta,x) = \\frac{1}{1+e^{-(\\theta_0 x_0 + \\theta_1 x_1 + \\theta_2)}}$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"$$\n",
"\\begin{array}{lcl}\n",
"f(x) = \\frac{1}{x} \\quad & \\rightarrow & \\quad \\frac{df}{dx} = -\\frac{1}{x^2} \\\\\n",
"f_c(x) = c + x \\quad & \\rightarrow & \\quad \\frac{df}{dx} = 1 \\\\\n",
"f(x) = e^x \\quad & \\rightarrow & \\quad \\frac{df}{dx} = e^x \\\\\n",
"f_a(x) = ax \\quad & \\rightarrow & \\quad \\frac{df}{dx} = a \\\\\n",
"\\end{array}\n",
"$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img src=\"exp2.png\" />"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[0.3932238664829637, -0.5898357997244456]\n",
"[-0.19661193324148185, -0.3932238664829637, 0.19661193324148185]\n"
]
}
],
"source": [
"# Losowe wagi i dane\n",
"w = [2,-3,-3]\n",
"x = [-1, -2]\n",
"\n",
"# Krok w przód\n",
"dot = w[0]*x[0] + w[1]*x[1] + w[2]\n",
"f = 1.0 / (1 + math.exp(-dot)) # funkcja sigmoidalna\n",
"\n",
"# Krok w tył\n",
"ddot = (1 - f) * f # pochodna funkcji sigmoidalnej\n",
"dx = [w[0] * ddot, w[1] * ddot]\n",
"dw = [x[0] * ddot, x[1] * ddot, 1.0 * ddot]\n",
"\n",
"print(dx)\n",
"print(dw)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Obliczanie gradientów –
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Gradient $f$ dla $x$ mówi, jak zmieni się całe wyrażenie przy zmianie wartości $x$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Gradienty łączymy, korzystając z **reguły łańcuchowej**."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* W kroku \"wstecz\" gradienty informują, które części grafu powinny być zwiększone lub zmniejszone (i z jaką siłą), aby zwiększyć wartość na wyjściu."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* W kontekście implementacji chcemy dzielić funkcję $f$ na części, dla których można łatwo obliczyć gradienty."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## 4.5. Uczenie wielowarstwowych sieci neuronowych metodą propagacji wstecznej"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"Mając algorytm SGD oraz gradienty wszystkich wag, moglibyśmy trenować każdą sieć."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Niech $\\Theta = (\\Theta^{(1)},\\Theta^{(2)},\\Theta^{(3)},\\beta^{(1)},\\beta^{(2)},\\beta^{(3)})$\n",
"* Funkcja sieci neuronowej z grafiki:\n",
"$$\\small h_\\Theta(x) = \\tanh(\\tanh(\\tanh(x\\Theta^{(1)}+\\beta^{(1)})\\Theta^{(2)} + \\beta^{(2)})\\Theta^{(3)} + \\beta^{(3)})$$\n",
"* Funkcja kosztu dla regresji:\n",
"$$J(\\Theta) = \\dfrac{1}{2m} \\sum_{i=1}^{m} (h_\\Theta(x^{(i)})- y^{(i)})^2 $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Jak obliczymy gradienty?\n",
"\n",
"$$\\nabla_{\\Theta^{(l)}} J(\\Theta) = ? \\quad \\nabla_{\\beta^{(l)}} J(\\Theta) = ?$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### W kierunku propagacji wstecznej\n",
"\n",
"* Pewna (niewielka) zmiana wagi $\\Delta z^l_j$ dla $j$-ego neuronu w warstwie $l$ pociąga za sobą (niewielką) zmianę kosztu: \n",
"\n",
"$$\\frac{\\partial J(\\Theta)}{\\partial z^{l}_j} \\Delta z^{l}_j$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Jeżeli $\\frac{\\partial J(\\Theta)}{\\partial z^{l}_j}$ jest duża, $\\Delta z^l_j$ ze znakiem przeciwnym zredukuje koszt.\n",
"* Jeżeli $\\frac{\\partial J(\\Theta)}{\\partial z^l_j}$ jest bliska zeru, koszt nie będzie mocno poprawiony."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"* Definiujemy błąd $\\delta^l_j$ neuronu $j$ w warstwie $l$: \n",
"\n",
"$$\\delta^l_j := \\dfrac{\\partial J(\\Theta)}{\\partial z^l_j}$$ \n",
"$$\\delta^l := \\nabla_{z^l} J(\\Theta) \\quad \\textrm{ (zapis wektorowy)} $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Podstawowe równania propagacji wstecznej\n",
"\n",
"$$\n",
"\\begin{array}{rcll}\n",
"\\delta^L & = & \\nabla_{a^L}J(\\Theta) \\odot { \\left( g^{L} \\right) }^{\\prime} \\left( z^L \\right) & (BP1) \\\\[2mm]\n",
"\\delta^{l} & = & \\left( \\left( \\Theta^{l+1} \\right) \\! ^\\top \\, \\delta^{l+1} \\right) \\odot {{ \\left( g^{l} \\right) }^{\\prime}} \\left( z^{l} \\right) & (BP2)\\\\[2mm]\n",
"\\nabla_{\\beta^l} J(\\Theta) & = & \\delta^l & (BP3)\\\\[2mm]\n",
"\\nabla_{\\Theta^l} J(\\Theta) & = & a^{l-1} \\odot \\delta^l & (BP4)\\\\\n",
"\\end{array}\n",
"$$\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### (BP1)\n",
"$$ \\delta^L_j \\; = \\; \\frac{ \\partial J }{ \\partial a^L_j } \\, g' \\!\\! \\left( z^L_j \\right) $$\n",
"$$ \\delta^L \\; = \\; \\nabla_{a^L}J(\\Theta) \\odot { \\left( g^{L} \\right) }^{\\prime} \\left( z^L \\right) $$\n",
"Błąd w ostatniej warstwie jest iloczynem szybkości zmiany kosztu względem $j$-tego wyjścia i szybkości zmiany funkcji aktywacji w punkcie $z^L_j$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### (BP2)\n",
"$$ \\delta^{l} \\; = \\; \\left( \\left( \\Theta^{l+1} \\right) \\! ^\\top \\, \\delta^{l+1} \\right) \\odot {{ \\left( g^{l} \\right) }^{\\prime}} \\left( z^{l} \\right) $$\n",
"Aby obliczyć błąd w $l$-tej warstwie, należy przemnożyć błąd z następnej ($(l+1)$-szej) warstwy przez transponowany wektor wag, a uzyskaną macierz pomnożyć po współrzędnych przez szybkość zmiany funkcji aktywacji w punkcie $z^l$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### (BP3)\n",
"$$ \\nabla_{\\beta^l} J(\\Theta) \\; = \\; \\delta^l $$\n",
"Błąd w $l$-tej warstwie jest równy wartości gradientu funkcji kosztu."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"#### (BP4)\n",
"$$ \\nabla_{\\Theta^l} J(\\Theta) \\; = \\; a^{l-1} \\odot \\delta^l $$\n",
"Gradient funkcji kosztu względem wag $l$-tej warstwy można obliczyć jako iloczyn po współrzędnych $a^{l-1}$ przez $\\delta^l$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Algorytm propagacji wstecznej"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"Dla pojedynczego przykładu $(x,y)$:\n",
"1. **Wejście**: Ustaw aktywacje w warstwie cech $a^{(0)}=x$ \n",
"2. **Feedforward:** dla $l=1,\\dots,L$ oblicz \n",
"$z^{(l)} = a^{(l-1)} \\Theta^{(l)} + \\beta^{(l)}$ oraz $a^{(l)}=g^{(l)} \\!\\! \\left( z^{(l)} \\right)$\n",
"3. **Błąd wyjścia $\\delta^{(L)}$:** oblicz wektor $$\\delta^{(L)}= \\nabla_{a^{(L)}}J(\\Theta) \\odot {g^{\\prime}}^{(L)} \\!\\! \\left( z^{(L)} \\right) $$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"4. **Propagacja wsteczna błędu:** dla $l = L-1,L-2,\\dots,1$ oblicz $$\\delta^{(l)} = \\delta^{(l+1)}(\\Theta^{(l+1)})^T \\odot {g^{\\prime}}^{(l)} \\!\\! \\left( z^{(l)} \\right) $$\n",
"5. **Gradienty:** \n",
" * $\\dfrac{\\partial}{\\partial \\Theta_{ij}^{(l)}} J(\\Theta) = a_i^{(l-1)}\\delta_j^{(l)} \\textrm{ oraz } \\dfrac{\\partial}{\\partial \\beta_{j}^{(l)}} J(\\Theta) = \\delta_j^{(l)}$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"W naszym przykładzie:\n",
"\n",
"$$\\small J(\\Theta) = \\frac{1}{2} \\left( a^{(L)} - y \\right) ^2 $$\n",
"$$\\small \\dfrac{\\partial}{\\partial a^{(L)}} J(\\Theta) = a^{(L)} - y$$\n",
"\n",
"$$\\small \\tanh^{\\prime}(x) = 1 - \\tanh^2(x)$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"<img src=\"nn3.png\" width=\"65%\"/>"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Algorytm SGD z propagacją wsteczną\n",
"\n",
"Pojedyncza iteracja:\n",
"1. Dla parametrów $\\Theta = (\\Theta^{(1)},\\ldots,\\Theta^{(L)})$ utwórz pomocnicze macierze zerowe $\\Delta = (\\Delta^{(1)},\\ldots,\\Delta^{(L)})$ o takich samych wymiarach (dla uproszczenia opuszczono wagi $\\beta$)."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"2. Dla $m$ przykładów we wsadzie (*batch*), $i = 1,\\ldots,m$:\n",
" * Wykonaj algortym propagacji wstecznej dla przykładu $(x^{(i)}, y^{(i)})$ i przechowaj gradienty $\\nabla_{\\Theta}J^{(i)}(\\Theta)$ dla tego przykładu;\n",
" * $\\Delta := \\Delta + \\dfrac{1}{m}\\nabla_{\\Theta}J^{(i)}(\\Theta)$\n",
"3. Wykonaj aktualizację
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Propagacja wsteczna –
"\n",
"* Algorytm pierwszy raz wprowadzony w latach 70. XX w.\n",
"* W 1986 David Rumelhart, Geoffrey Hinton i Ronald Williams pokazali, że jest znacznie szybszy od wcześniejszych metod.\n",
"* Obecnie najpopularniejszy algorytm uczenia sieci neuronowych."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
2021-05-19 18:55:21 +02:00
"## 4.6. Przykłady implementacji wielowarstwowych sieci neuronowych"
2021-05-05 15:23:24 +02:00
]
},
{
2021-05-19 18:55:21 +02:00
"cell_type": "markdown",
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
2021-05-19 18:55:21 +02:00
"slide_type": "notes"
2021-05-05 15:23:24 +02:00
}
},
"source": [
2021-05-19 18:55:21 +02:00
"### Uwaga!\n",
2021-05-05 15:23:24 +02:00
"\n",
2021-05-19 18:55:21 +02:00
"Poniższe przykłady wykorzystują interfejs [Keras](https://keras.io), który jest częścią biblioteki [TensorFlow](https://www.tensorflow.org).\n",
2021-05-05 15:23:24 +02:00
"\n",
2021-05-19 18:55:21 +02:00
"Aby uruchomić TensorFlow w środowisku Jupyter, należy wykonać następujące czynności:\n",
2021-05-05 15:23:24 +02:00
"\n",
2021-05-19 18:55:21 +02:00
"#### Przed pierwszym uruchomieniem (wystarczy wykonać tylko raz)\n",
2021-05-05 15:23:24 +02:00
"\n",
2021-05-19 18:55:21 +02:00
"Instalacja biblioteki TensorFlow w środowisku Anaconda:\n",
"\n",
"1. Uruchom *Anaconda Navigator*\n",
"1. Wybierz kafelek *CMD.exe Prompt*\n",
"1. Kliknij przycisk *Launch*\n",
"1. Pojawi się konsola. Wpisz następujące polecenia, każde zatwierdzając wciśnięciem klawisza Enter:\n",
"```\n",
"conda create -n tf tensorflow\n",
"conda activate tf\n",
"conda install pandas matplotlib\n",
"jupyter notebook\n",
"```\n",
"\n",
"#### Przed każdym uruchomieniem\n",
"\n",
"Jeżeli chcemy korzystać z biblioteki TensorFlow, to środowisko Jupyter Notebook należy uruchomić w następujący sposób:\n",
"\n",
"1. Uruchom *Anaconda Navigator*\n",
"1. Wybierz kafelek *CMD.exe Prompt*\n",
"1. Kliknij przycisk *Launch*\n",
"1. Pojawi się konsola. Wpisz następujące polecenia, każde zatwierdzając wciśnięciem klawisza Enter:\n",
"```\n",
"conda activate tf\n",
"jupyter notebook\n",
"```"
2021-05-05 15:23:24 +02:00
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"### Przykład: MNIST\n",
"\n",
"_Modified National Institute of Standards and Technology database_"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"* Zbiór cyfr zapisanych pismem odręcznym\n",
"* 60 000 przykładów uczących, 10 000 przykładów testowych\n",
"* Rozdzielczość każdego przykładu: 28 ×
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 52,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [],
"source": [
"# źródło: https://github.com/keras-team/keras/examples/minst_mlp.py\n",
"\n",
2021-05-19 18:55:21 +02:00
"from tensorflow import keras\n",
"from tensorflow.keras.datasets import mnist\n",
"from tensorflow.keras.layers import Dense, Dropout\n",
2021-05-05 15:23:24 +02:00
"\n",
"# załaduj dane i podziel je na zbiory uczący i testowy\n",
"(x_train, y_train), (x_test, y_test) = mnist.load_data()"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 53,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [],
"source": [
2021-05-19 18:55:21 +02:00
"from matplotlib import pyplot as plt\n",
"\n",
2021-05-05 15:23:24 +02:00
"def draw_examples(examples, captions=None):\n",
" plt.figure(figsize=(16, 4))\n",
" m = len(examples)\n",
" for i, example in enumerate(examples):\n",
" plt.subplot(100 + m * 10 + i + 1)\n",
" plt.imshow(example, cmap=plt.get_cmap('gray'))\n",
" plt.show()\n",
" if captions is not None:\n",
" print(6 * ' ' + (10 * ' ').join(str(captions[i]) for i in range(m)))"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 54,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"data": {
2021-05-19 18:55:21 +02:00
"image/png": "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
2021-05-05 15:23:24 +02:00
"text/plain": [
2021-05-19 18:55:21 +02:00
"<Figure size 1152x288 with 7 Axes>"
2021-05-05 15:23:24 +02:00
]
},
2021-05-19 18:55:21 +02:00
"metadata": {
"needs_background": "light"
},
2021-05-05 15:23:24 +02:00
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" 5 0 4 1 9 2 1\n"
]
}
],
"source": [
"draw_examples(x_train[:7], captions=y_train)"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 55,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"60000 przykładów uczących\n",
"10000 przykładów testowych\n"
]
}
],
"source": [
"num_classes = 10\n",
"\n",
"x_train = x_train.reshape(60000, 784) # 784 = 28 * 28\n",
"x_test = x_test.reshape(10000, 784)\n",
"x_train = x_train.astype('float32')\n",
"x_test = x_test.astype('float32')\n",
"x_train /= 255\n",
"x_test /= 255\n",
"print('{} przykładów uczących'.format(x_train.shape[0]))\n",
"print('{} przykładów testowych'.format(x_test.shape[0]))\n",
"\n",
"# przekonwertuj wektory klas na binarne macierze klas\n",
"y_train = keras.utils.to_categorical(y_train, num_classes)\n",
"y_test = keras.utils.to_categorical(y_test, num_classes)"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 56,
2021-05-05 15:23:24 +02:00
"metadata": {
"scrolled": true,
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Model: \"sequential_21\"\n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
2021-05-19 18:55:21 +02:00
"dense_59 (Dense) (None, 512) 401920 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dropout_2 (Dropout) (None, 512) 0 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_60 (Dense) (None, 512) 262656 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dropout_3 (Dropout) (None, 512) 0 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_61 (Dense) (None, 10) 5130 \n",
2021-05-05 15:23:24 +02:00
"=================================================================\n",
"Total params: 669,706\n",
"Trainable params: 669,706\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
]
}
],
"source": [
2021-05-19 18:55:21 +02:00
"model = keras.Sequential()\n",
2021-05-05 15:23:24 +02:00
"model.add(Dense(512, activation='relu', input_shape=(784,)))\n",
"model.add(Dropout(0.2))\n",
"model.add(Dense(512, activation='relu'))\n",
"model.add(Dropout(0.2))\n",
"model.add(Dense(num_classes, activation='softmax'))\n",
"model.summary()"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 57,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"(60000, 784) (60000, 10)\n"
2021-05-05 15:23:24 +02:00
]
}
],
"source": [
"print(x_train.shape, y_train.shape)"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 58,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 11s 23ms/step - loss: 0.2463 - accuracy: 0.9238 - val_loss: 0.1009 - val_accuracy: 0.9690\n",
2021-05-05 15:23:24 +02:00
"Epoch 2/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 10s 22ms/step - loss: 0.1042 - accuracy: 0.9681 - val_loss: 0.0910 - val_accuracy: 0.9739\n",
2021-05-05 15:23:24 +02:00
"Epoch 3/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 11s 23ms/step - loss: 0.0774 - accuracy: 0.9762 - val_loss: 0.0843 - val_accuracy: 0.9755\n",
2021-05-05 15:23:24 +02:00
"Epoch 4/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 11s 24ms/step - loss: 0.0606 - accuracy: 0.9815 - val_loss: 0.0691 - val_accuracy: 0.9818\n",
2021-05-05 15:23:24 +02:00
"Epoch 5/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 10s 22ms/step - loss: 0.0504 - accuracy: 0.9848 - val_loss: 0.0886 - val_accuracy: 0.9772\n"
2021-05-05 15:23:24 +02:00
]
},
{
"data": {
"text/plain": [
2021-05-19 18:55:21 +02:00
"<tensorflow.python.keras.callbacks.History at 0x1ed6e0565b0>"
2021-05-05 15:23:24 +02:00
]
},
2021-05-19 18:55:21 +02:00
"execution_count": 58,
2021-05-05 15:23:24 +02:00
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
2021-05-19 18:55:21 +02:00
"model.compile(loss='categorical_crossentropy', optimizer=keras.optimizers.RMSprop(), metrics=['accuracy'])\n",
2021-05-05 15:23:24 +02:00
"\n",
"model.fit(x_train, y_train, batch_size=128, epochs=5, verbose=1,\n",
" validation_data=(x_test, y_test))"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 60,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Test loss: 0.08859136700630188\n",
"Test accuracy: 0.9771999716758728\n"
2021-05-05 15:23:24 +02:00
]
}
],
"source": [
"score = model.evaluate(x_test, y_test, verbose=0)\n",
"\n",
"print('Test loss: {}'.format(score[0]))\n",
"print('Test accuracy: {}'.format(score[1]))"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"source": [
"Warstwa _dropout_ to metoda regularyzacji, służy zapobieganiu nadmiernemu dopasowaniu sieci. Polega na tym, że część węzłów sieci jest usuwana w sposób losowy."
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 61,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Model: \"sequential_22\"\n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
2021-05-19 18:55:21 +02:00
"dense_62 (Dense) (None, 512) 401920 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_63 (Dense) (None, 512) 262656 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_64 (Dense) (None, 10) 5130 \n",
2021-05-05 15:23:24 +02:00
"=================================================================\n",
"Total params: 669,706\n",
"Trainable params: 669,706\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n",
"Epoch 1/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 10s 20ms/step - loss: 0.2203 - accuracy: 0.9317 - val_loss: 0.0936 - val_accuracy: 0.9697\n",
2021-05-05 15:23:24 +02:00
"Epoch 2/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 10s 21ms/step - loss: 0.0816 - accuracy: 0.9746 - val_loss: 0.0747 - val_accuracy: 0.9779\n",
2021-05-05 15:23:24 +02:00
"Epoch 3/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 10s 20ms/step - loss: 0.0544 - accuracy: 0.9827 - val_loss: 0.0674 - val_accuracy: 0.9798\n",
2021-05-05 15:23:24 +02:00
"Epoch 4/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 10s 22ms/step - loss: 0.0384 - accuracy: 0.9879 - val_loss: 0.0746 - val_accuracy: 0.9806\n",
2021-05-05 15:23:24 +02:00
"Epoch 5/5\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 10s 22ms/step - loss: 0.0298 - accuracy: 0.9901 - val_loss: 0.0736 - val_accuracy: 0.9801\n"
2021-05-05 15:23:24 +02:00
]
},
{
"data": {
"text/plain": [
2021-05-19 18:55:21 +02:00
"<tensorflow.python.keras.callbacks.History at 0x1ed7eba8070>"
2021-05-05 15:23:24 +02:00
]
},
2021-05-19 18:55:21 +02:00
"execution_count": 61,
2021-05-05 15:23:24 +02:00
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Bez warstw Dropout\n",
"\n",
"num_classes = 10\n",
"\n",
"(x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
"\n",
"x_train = x_train.reshape(60000, 784) # 784 = 28 * 28\n",
"x_test = x_test.reshape(10000, 784)\n",
"x_train = x_train.astype('float32')\n",
"x_test = x_test.astype('float32')\n",
"x_train /= 255\n",
"x_test /= 255\n",
"\n",
"y_train = keras.utils.to_categorical(y_train, num_classes)\n",
"y_test = keras.utils.to_categorical(y_test, num_classes)\n",
"\n",
2021-05-19 18:55:21 +02:00
"model_no_dropout = keras.Sequential()\n",
2021-05-05 15:23:24 +02:00
"model_no_dropout.add(Dense(512, activation='relu', input_shape=(784,)))\n",
"model_no_dropout.add(Dense(512, activation='relu'))\n",
"model_no_dropout.add(Dense(num_classes, activation='softmax'))\n",
"model_no_dropout.summary()\n",
"\n",
"model_no_dropout.compile(loss='categorical_crossentropy',\n",
2021-05-19 18:55:21 +02:00
" optimizer=keras.optimizers.RMSprop(),\n",
2021-05-05 15:23:24 +02:00
" metrics=['accuracy'])\n",
"\n",
"model_no_dropout.fit(x_train, y_train,\n",
" batch_size=128,\n",
" epochs=5,\n",
" verbose=1,\n",
" validation_data=(x_test, y_test))"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 62,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Test loss (no dropout): 0.07358124107122421\n",
"Test accuracy (no dropout): 0.9800999760627747\n"
2021-05-05 15:23:24 +02:00
]
}
],
"source": [
"# Bez warstw Dropout\n",
"\n",
"score = model_no_dropout.evaluate(x_test, y_test, verbose=0)\n",
"\n",
"print('Test loss (no dropout): {}'.format(score[0]))\n",
"print('Test accuracy (no dropout): {}'.format(score[1]))"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 63,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Model: \"sequential_23\"\n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
2021-05-19 18:55:21 +02:00
"dense_65 (Dense) (None, 2500) 1962500 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_66 (Dense) (None, 2000) 5002000 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_67 (Dense) (None, 1500) 3001500 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_68 (Dense) (None, 1000) 1501000 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_69 (Dense) (None, 500) 500500 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_70 (Dense) (None, 10) 5010 \n",
2021-05-05 15:23:24 +02:00
"=================================================================\n",
"Total params: 11,972,510\n",
"Trainable params: 11,972,510\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n",
"Epoch 1/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 129s 275ms/step - loss: 0.9587 - accuracy: 0.7005 - val_loss: 0.5066 - val_accuracy: 0.8566\n",
2021-05-05 15:23:24 +02:00
"Epoch 2/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 130s 276ms/step - loss: 0.2666 - accuracy: 0.9234 - val_loss: 0.3376 - val_accuracy: 0.9024\n",
2021-05-05 15:23:24 +02:00
"Epoch 3/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 130s 277ms/step - loss: 0.1811 - accuracy: 0.9477 - val_loss: 0.1678 - val_accuracy: 0.9520\n",
2021-05-05 15:23:24 +02:00
"Epoch 4/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 134s 287ms/step - loss: 0.1402 - accuracy: 0.9588 - val_loss: 0.1553 - val_accuracy: 0.9576\n",
2021-05-05 15:23:24 +02:00
"Epoch 5/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 130s 278ms/step - loss: 0.1153 - accuracy: 0.9662 - val_loss: 0.1399 - val_accuracy: 0.9599\n",
2021-05-05 15:23:24 +02:00
"Epoch 6/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 130s 277ms/step - loss: 0.0956 - accuracy: 0.9711 - val_loss: 0.1389 - val_accuracy: 0.9612\n",
2021-05-05 15:23:24 +02:00
"Epoch 7/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 131s 280ms/step - loss: 0.0803 - accuracy: 0.9761 - val_loss: 0.1008 - val_accuracy: 0.9724\n",
2021-05-05 15:23:24 +02:00
"Epoch 8/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 134s 286ms/step - loss: 0.0685 - accuracy: 0.9797 - val_loss: 0.1137 - val_accuracy: 0.9679\n",
2021-05-05 15:23:24 +02:00
"Epoch 9/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 130s 278ms/step - loss: 0.0602 - accuracy: 0.9819 - val_loss: 0.1064 - val_accuracy: 0.9700\n",
2021-05-05 15:23:24 +02:00
"Epoch 10/10\n",
2021-05-19 18:55:21 +02:00
"469/469 [==============================] - 129s 274ms/step - loss: 0.0520 - accuracy: 0.9843 - val_loss: 0.1095 - val_accuracy: 0.9698\n"
2021-05-05 15:23:24 +02:00
]
},
{
"data": {
"text/plain": [
2021-05-19 18:55:21 +02:00
"<tensorflow.python.keras.callbacks.History at 0x1ed0e628250>"
2021-05-05 15:23:24 +02:00
]
},
2021-05-19 18:55:21 +02:00
"execution_count": 63,
2021-05-05 15:23:24 +02:00
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Więcej warstw, inna funkcja aktywacji\n",
"\n",
"num_classes = 10\n",
"\n",
"(x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
"\n",
"x_train = x_train.reshape(60000, 784) # 784 = 28 * 28\n",
"x_test = x_test.reshape(10000, 784)\n",
"x_train = x_train.astype('float32')\n",
"x_test = x_test.astype('float32')\n",
"x_train /= 255\n",
"x_test /= 255\n",
"\n",
"y_train = keras.utils.to_categorical(y_train, num_classes)\n",
"y_test = keras.utils.to_categorical(y_test, num_classes)\n",
"\n",
2021-06-07 14:50:30 +02:00
"model3 = keras.Sequential()\n",
2021-05-05 15:23:24 +02:00
"model3.add(Dense(2500, activation='tanh', input_shape=(784,)))\n",
"model3.add(Dense(2000, activation='tanh'))\n",
"model3.add(Dense(1500, activation='tanh'))\n",
"model3.add(Dense(1000, activation='tanh'))\n",
"model3.add(Dense(500, activation='tanh'))\n",
"model3.add(Dense(num_classes, activation='softmax'))\n",
"model3.summary()\n",
"\n",
"model3.compile(loss='categorical_crossentropy',\n",
2021-05-19 18:55:21 +02:00
" optimizer=keras.optimizers.RMSprop(),\n",
2021-05-05 15:23:24 +02:00
" metrics=['accuracy'])\n",
"\n",
"model3.fit(x_train, y_train,\n",
" batch_size=128,\n",
" epochs=10,\n",
" verbose=1,\n",
" validation_data=(x_test, y_test))"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 64,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Test loss: 0.10945799201726913\n",
"Test accuracy: 0.9697999954223633\n"
2021-05-05 15:23:24 +02:00
]
}
],
"source": [
"# Więcej warstw, inna funkcja aktywacji\n",
"\n",
"score = model3.evaluate(x_test, y_test, verbose=0)\n",
"\n",
"print('Test loss: {}'.format(score[0]))\n",
"print('Test accuracy: {}'.format(score[1]))"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Przykład: 4-pikselowy aparat fotograficzny\n",
"\n",
"https://www.youtube.com/watch?v=ILsA4nyG7I0"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 65,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [],
"source": [
"def generate_example(description):\n",
" variant = random.choice([1, -1])\n",
" if description == 's': # solid\n",
" return (np.array([[ 1.0, 1.0], [ 1.0, 1.0]]) if variant == 1 else\n",
" np.array([[-1.0, -1.0], [-1.0, -1.0]]))\n",
" elif description == 'v': # vertical\n",
" return (np.array([[ 1.0, -1.0], [ 1.0, -1.0]]) if variant == 1 else\n",
" np.array([[-1.0, 1.0], [-1.0, 1.0]]))\n",
" elif description == 'd': # diagonal\n",
" return (np.array([[ 1.0, -1.0], [-1.0, 1.0]]) if variant == 1 else\n",
" np.array([[-1.0, 1.0], [ 1.0, -1.0]]))\n",
" elif description == 'h': # horizontal\n",
" return (np.array([[ 1.0, 1.0], [-1.0, -1.0]]) if variant == 1 else\n",
" np.array([[-1.0, -1.0], [ 1.0, 1.0]]))\n",
" else:\n",
" return np.array([[random.uniform(-1, 1), random.uniform(-1, 1)],\n",
" [random.uniform(-1, 1), random.uniform(-1, 1)]])"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 67,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [],
"source": [
2021-05-19 18:55:21 +02:00
"import random\n",
"\n",
2021-05-05 15:23:24 +02:00
"num_classes = 4\n",
"\n",
"trainset_size = 4000\n",
"testset_size = 1000\n",
"\n",
"y4_train = np.array([random.choice(['s', 'v', 'd', 'h']) for i in range(trainset_size)])\n",
"x4_train = np.array([generate_example(desc) for desc in y4_train])\n",
"\n",
"y4_test = np.array([random.choice(['s', 'v', 'd', 'h']) for i in range(testset_size)])\n",
"x4_test = np.array([generate_example(desc) for desc in y4_test])"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 68,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"data": {
2021-05-19 18:55:21 +02:00
"image/png": "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
2021-05-05 15:23:24 +02:00
"text/plain": [
2021-05-19 18:55:21 +02:00
"<Figure size 1152x288 with 7 Axes>"
2021-05-05 15:23:24 +02:00
]
},
2021-05-19 18:55:21 +02:00
"metadata": {
"needs_background": "light"
},
2021-05-05 15:23:24 +02:00
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
" s d h s d v v\n"
2021-05-05 15:23:24 +02:00
]
}
],
"source": [
"draw_examples(x4_train[:7], captions=y4_train)"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 69,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [],
"source": [
"x4_train = x4_train.reshape(trainset_size, 4)\n",
"x4_test = x4_test.reshape(testset_size, 4)\n",
"x4_train = x4_train.astype('float32')\n",
"x4_test = x4_test.astype('float32')\n",
"\n",
"y4_train = np.array([{'s': 0, 'v': 1, 'd': 2, 'h': 3}[desc] for desc in y4_train])\n",
"y4_test = np.array([{'s': 0, 'v': 1, 'd': 2, 'h': 3}[desc] for desc in y4_test])\n",
"\n",
"y4_train = keras.utils.to_categorical(y4_train, num_classes)\n",
"y4_test = keras.utils.to_categorical(y4_test, num_classes)"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 70,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Model: \"sequential_24\"\n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
2021-05-19 18:55:21 +02:00
"dense_71 (Dense) (None, 4) 20 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_72 (Dense) (None, 4) 20 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_73 (Dense) (None, 8) 40 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_74 (Dense) (None, 4) 36 \n",
2021-05-05 15:23:24 +02:00
"=================================================================\n",
"Total params: 116\n",
"Trainable params: 116\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
]
}
],
"source": [
2021-05-19 18:55:21 +02:00
"model4 = keras.Sequential()\n",
2021-05-05 15:23:24 +02:00
"model4.add(Dense(4, activation='tanh', input_shape=(4,)))\n",
"model4.add(Dense(4, activation='tanh'))\n",
"model4.add(Dense(8, activation='relu'))\n",
"model4.add(Dense(num_classes, activation='softmax'))\n",
"model4.summary()"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 71,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [],
"source": [
"model4.layers[0].set_weights(\n",
" [np.array([[ 1.0, 0.0, 1.0, 0.0],\n",
" [ 0.0, 1.0, 0.0, 1.0],\n",
" [ 1.0, 0.0, -1.0, 0.0],\n",
" [ 0.0, 1.0, 0.0, -1.0]],\n",
" dtype=np.float32), np.array([0., 0., 0., 0.], dtype=np.float32)])\n",
"model4.layers[1].set_weights(\n",
" [np.array([[ 1.0, -1.0, 0.0, 0.0],\n",
" [ 1.0, 1.0, 0.0, 0.0],\n",
" [ 0.0, 0.0, 1.0, -1.0],\n",
" [ 0.0, 0.0, -1.0, -1.0]],\n",
" dtype=np.float32), np.array([0., 0., 0., 0.], dtype=np.float32)])\n",
"model4.layers[2].set_weights(\n",
" [np.array([[ 1.0, -1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],\n",
" [ 0.0, 0.0, 1.0, -1.0, 0.0, 0.0, 0.0, 0.0],\n",
" [ 0.0, 0.0, 0.0, 0.0, 1.0, -1.0, 0.0, 0.0],\n",
" [ 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, -1.0]],\n",
" dtype=np.float32), np.array([0., 0., 0., 0., 0., 0., 0., 0.], dtype=np.float32)])"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 73,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [],
"source": [
"model4.layers[3].set_weights(\n",
" [np.array([[ 1.0, 0.0, 0.0, 0.0],\n",
" [ 1.0, 0.0, 0.0, 0.0],\n",
" [ 0.0, 1.0, 0.0, 0.0],\n",
" [ 0.0, 1.0, 0.0, 0.0],\n",
" [ 0.0, 0.0, 1.0, 0.0],\n",
" [ 0.0, 0.0, 1.0, 0.0],\n",
" [ 0.0, 0.0, 0.0, 1.0],\n",
" [ 0.0, 0.0, 0.0, 1.0]],\n",
" dtype=np.float32), np.array([0., 0., 0., 0.], dtype=np.float32)])\n",
"\n",
"model4.compile(loss='categorical_crossentropy',\n",
2021-05-19 18:55:21 +02:00
" optimizer=keras.optimizers.Adagrad(),\n",
2021-05-05 15:23:24 +02:00
" metrics=['accuracy'])"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 74,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[array([[ 1., 0., 1., 0.],\n",
" [ 0., 1., 0., 1.],\n",
" [ 1., 0., -1., 0.],\n",
2021-05-19 18:55:21 +02:00
" [ 0., 1., 0., -1.]], dtype=float32), array([0., 0., 0., 0.], dtype=float32)]\n",
2021-05-05 15:23:24 +02:00
"[array([[ 1., -1., 0., 0.],\n",
" [ 1., 1., 0., 0.],\n",
" [ 0., 0., 1., -1.],\n",
2021-05-19 18:55:21 +02:00
" [ 0., 0., -1., -1.]], dtype=float32), array([0., 0., 0., 0.], dtype=float32)]\n",
2021-05-05 15:23:24 +02:00
"[array([[ 1., -1., 0., 0., 0., 0., 0., 0.],\n",
" [ 0., 0., 1., -1., 0., 0., 0., 0.],\n",
" [ 0., 0., 0., 0., 1., -1., 0., 0.],\n",
2021-05-19 18:55:21 +02:00
" [ 0., 0., 0., 0., 0., 0., 1., -1.]], dtype=float32), array([0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32)]\n",
"[array([[1., 0., 0., 0.],\n",
" [1., 0., 0., 0.],\n",
" [0., 1., 0., 0.],\n",
" [0., 1., 0., 0.],\n",
" [0., 0., 1., 0.],\n",
" [0., 0., 1., 0.],\n",
" [0., 0., 0., 1.],\n",
" [0., 0., 0., 1.]], dtype=float32), array([0., 0., 0., 0.], dtype=float32)]\n"
2021-05-05 15:23:24 +02:00
]
}
],
"source": [
"for layer in model4.layers:\n",
" print(layer.get_weights())"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 75,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"data": {
"text/plain": [
2021-05-19 18:55:21 +02:00
"array([[0.17831734, 0.17831734, 0.17831734, 0.465048 ]], dtype=float32)"
2021-05-05 15:23:24 +02:00
]
},
2021-05-19 18:55:21 +02:00
"execution_count": 75,
2021-05-05 15:23:24 +02:00
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"model4.predict([np.array([[1.0, 1.0], [-1.0, -1.0]]).reshape(1, 4)])"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 76,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Test loss: 0.7656148672103882\n",
2021-05-05 15:23:24 +02:00
"Test accuracy: 1.0\n"
]
}
],
"source": [
"score = model4.evaluate(x4_test, y4_test, verbose=0)\n",
"\n",
"print('Test loss: {}'.format(score[0]))\n",
"print('Test accuracy: {}'.format(score[1]))"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 77,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Model: \"sequential_25\"\n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
2021-05-19 18:55:21 +02:00
"dense_75 (Dense) (None, 4) 20 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_76 (Dense) (None, 4) 20 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_77 (Dense) (None, 8) 40 \n",
2021-05-05 15:23:24 +02:00
"_________________________________________________________________\n",
2021-05-19 18:55:21 +02:00
"dense_78 (Dense) (None, 4) 36 \n",
2021-05-05 15:23:24 +02:00
"=================================================================\n",
"Total params: 116\n",
"Trainable params: 116\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
]
}
],
"source": [
2021-06-07 14:50:30 +02:00
"model5 = keras.Sequential()\n",
2021-05-05 15:23:24 +02:00
"model5.add(Dense(4, activation='tanh', input_shape=(4,)))\n",
"model5.add(Dense(4, activation='tanh'))\n",
"model5.add(Dense(8, activation='relu'))\n",
"model5.add(Dense(num_classes, activation='softmax'))\n",
"model5.compile(loss='categorical_crossentropy',\n",
2021-05-19 18:55:21 +02:00
" optimizer=keras.optimizers.RMSprop(),\n",
2021-05-05 15:23:24 +02:00
" metrics=['accuracy'])\n",
"model5.summary()"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 78,
2021-05-05 15:23:24 +02:00
"metadata": {
"scrolled": true,
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/8\n",
2021-05-19 18:55:21 +02:00
"125/125 [==============================] - 0s 3ms/step - loss: 1.3126 - accuracy: 0.3840 - val_loss: 1.1926 - val_accuracy: 0.6110\n",
2021-05-05 15:23:24 +02:00
"Epoch 2/8\n",
2021-05-19 18:55:21 +02:00
"125/125 [==============================] - 0s 2ms/step - loss: 1.0978 - accuracy: 0.5980 - val_loss: 1.0085 - val_accuracy: 0.6150\n",
2021-05-05 15:23:24 +02:00
"Epoch 3/8\n",
2021-05-19 18:55:21 +02:00
"125/125 [==============================] - 0s 2ms/step - loss: 0.9243 - accuracy: 0.7035 - val_loss: 0.8416 - val_accuracy: 0.7380\n",
2021-05-05 15:23:24 +02:00
"Epoch 4/8\n",
2021-05-19 18:55:21 +02:00
"125/125 [==============================] - 0s 2ms/step - loss: 0.7522 - accuracy: 0.8740 - val_loss: 0.6738 - val_accuracy: 1.0000\n",
2021-05-05 15:23:24 +02:00
"Epoch 5/8\n",
2021-05-19 18:55:21 +02:00
"125/125 [==============================] - 0s 2ms/step - loss: 0.5811 - accuracy: 1.0000 - val_loss: 0.5030 - val_accuracy: 1.0000\n",
2021-05-05 15:23:24 +02:00
"Epoch 6/8\n",
2021-05-19 18:55:21 +02:00
"125/125 [==============================] - 0s 2ms/step - loss: 0.4134 - accuracy: 1.0000 - val_loss: 0.3428 - val_accuracy: 1.0000\n",
2021-05-05 15:23:24 +02:00
"Epoch 7/8\n",
2021-05-19 18:55:21 +02:00
"125/125 [==============================] - 0s 2ms/step - loss: 0.2713 - accuracy: 1.0000 - val_loss: 0.2161 - val_accuracy: 1.0000\n",
2021-05-05 15:23:24 +02:00
"Epoch 8/8\n",
2021-05-19 18:55:21 +02:00
"125/125 [==============================] - 0s 1ms/step - loss: 0.1621 - accuracy: 1.0000 - val_loss: 0.1225 - val_accuracy: 1.0000\n"
2021-05-05 15:23:24 +02:00
]
},
{
"data": {
"text/plain": [
2021-05-19 18:55:21 +02:00
"<tensorflow.python.keras.callbacks.History at 0x1ed00809700>"
2021-05-05 15:23:24 +02:00
]
},
2021-05-19 18:55:21 +02:00
"execution_count": 78,
2021-05-05 15:23:24 +02:00
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"model5.fit(x4_train, y4_train, epochs=8, validation_data=(x4_test, y4_test))"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 79,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"data": {
"text/plain": [
2021-05-19 18:55:21 +02:00
"array([[3.2040708e-02, 1.0065207e-03, 4.9596769e-04, 9.6645677e-01]],\n",
" dtype=float32)"
2021-05-05 15:23:24 +02:00
]
},
2021-05-19 18:55:21 +02:00
"execution_count": 79,
2021-05-05 15:23:24 +02:00
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"model5.predict([np.array([[1.0, 1.0], [-1.0, -1.0]]).reshape(1, 4)])"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 80,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"Test loss: 0.1224619448184967\n",
2021-05-05 15:23:24 +02:00
"Test accuracy: 1.0\n"
]
}
],
"source": [
"score = model5.evaluate(x4_test, y4_test, verbose=0)\n",
"\n",
"print('Test loss: {}'.format(score[0]))\n",
"print('Test accuracy: {}'.format(score[1]))"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 81,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "notes"
}
},
"outputs": [],
"source": [
"import contextlib\n",
"\n",
"@contextlib.contextmanager\n",
"def printoptions(*args, **kwargs):\n",
" original = np.get_printoptions()\n",
" np.set_printoptions(*args, **kwargs)\n",
" try:\n",
" yield\n",
" finally: \n",
" np.set_printoptions(**original)"
]
},
{
"cell_type": "code",
2021-05-19 18:55:21 +02:00
"execution_count": 82,
2021-05-05 15:23:24 +02:00
"metadata": {
"slideshow": {
"slide_type": "subslide"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
2021-05-19 18:55:21 +02:00
"[array([[ 0.7, 0.2, -0.7, 0.7],\n",
" [-0.5, 0.9, 0.6, 0.6],\n",
" [ 1.1, 0.2, 0.1, 0.2],\n",
" [ 0.7, 0.1, 0.3, -0.7]], dtype=float32), array([ 0. , 0.1, -0.1, -0.2], dtype=float32)]\n",
"[array([[ 0.7, 0.5, -1.1, -1.2],\n",
" [ 0.7, 0.9, -0.6, 0.3],\n",
" [ 0.1, 1.4, -0.6, 0.8],\n",
" [ 1.5, 0.1, -0.1, 0.9]], dtype=float32), array([-0.4, 0.2, -0. , 0.2], dtype=float32)]\n",
"[array([[-1. , 1. , -0.7, -0.3, 0.2, 1.3, -0.7, 0.9],\n",
" [-0.9, 0.5, 0.8, -1.3, -1.2, 1.3, 0.4, -1. ],\n",
" [ 0.9, 0.2, 0.3, 0.4, 1.3, -0.9, -0.1, -0.2],\n",
" [-0.4, 0.5, 1.1, -0.6, 1.1, 0.1, -1.5, -1. ]], dtype=float32), array([-0.1, 0.1, 0.1, 0.1, 0.2, -0. , 0.1, 0.2], dtype=float32)]\n",
"[array([[ 0.7, -0.5, 0.8, -0.5],\n",
" [-0.3, -1.6, -0.2, 0.1],\n",
" [-1.5, 0.9, 0.1, -0.5],\n",
" [ 0.6, 0.7, 1. , -1.4],\n",
" [ 0.7, -1.2, -1.6, 1.2],\n",
" [ 1. , -1.2, 0.3, -1.5],\n",
" [-0.2, 0. , 0.6, 1.3],\n",
" [-0.8, 0.2, -0.6, -1. ]], dtype=float32), array([-0.6, 0.5, -0.3, 0.4], dtype=float32)]\n"
2021-05-05 15:23:24 +02:00
]
}
],
"source": [
"with printoptions(precision=1, suppress=True):\n",
" for layer in model5.layers:\n",
" print(layer.get_weights())"
]
2021-05-05 14:43:56 +02:00
}
],
"metadata": {
"celltoolbar": "Slideshow",
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
2021-06-07 14:50:30 +02:00
"version": "3.8.3"
2021-05-05 14:43:56 +02:00
},
"livereveal": {
"start_slideshow_at": "selected",
"theme": "white"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
