tutorials/5-Classification-with-RC.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Classification with Reservoir Computing\n",
    "\n",
    "Reservoir Computing (RC) is well suited to both regression and classification tasks. In the following notebook, you will experiment with a simple example of classification task."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from collections import defaultdict\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "from reservoirpy.datasets import japanese_vowels\n",
    "from reservoirpy import set_seed, verbosity\n",
    "from reservoirpy.observables import nrmse, rsquare\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "set_seed(42)\n",
    "verbosity(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Classification - The Japanese vowel dataset\n",
    "\n",
    "The Japanese vowel dataset is composed of 640 utterances of the Japanese vowel `\\ae\\`, from 9 different male speakers. The goal of this task is to assign to each utterance the label of its speaker. Dataset is split between a 270 utterances training set and a 340 utterances testing set.\n",
    "\n",
    "Each spoken utterance is a timeseries of 7~29 timesteps. Each timestep of signal is a 12 dimensional vector representing Linear Prediction Coefficient (LPC), which encode the audio signal into the cepstral domain (a variant of the frequency domain).\n",
    "\n",
    "\n",
    "### References\n",
    "\n",
    "M. Kudo, J. Toyama and M. Shimbo. (1999). \"Multidimensional Curve Classification Using Passing-Through Regions\". Pattern Recognition Letters, Vol. 20, No. 11--13, pages 1103--1111.\n",
    "\n",
    "https://archive.ics.uci.edu/ml/machine-learning-databases/JapaneseVowels-mld/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "X_train, Y_train, X_test, Y_test = japanese_vowels()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.imshow(X_train[0].T, vmin=-1.2, vmax=2)\n",
    "plt.title(f\"A sample vowel of speaker {np.argmax(Y_train[0]) +1}\")\n",
    "plt.xlabel(\"Timesteps\")\n",
    "plt.ylabel(\"LPC (cepstra)\")\n",
    "plt.colorbar()\n",
    "plt.show()\n",
    "\n",
    "plt.figure()\n",
    "plt.imshow(X_train[50].T, vmin=-1.2, vmax=2)\n",
    "plt.title(f\"A sample vowel of speaker {np.argmax(Y_train[50]) +1}\")\n",
    "plt.xlabel(\"Timesteps\")\n",
    "plt.ylabel(\"LPC (cepstra)\")\n",
    "plt.colorbar()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sample_per_speaker = 30\n",
    "n_speaker = 9\n",
    "X_train_per_speaker = []\n",
    "\n",
    "for i in range(n_speaker):\n",
    "    X_speaker = X_train[i*sample_per_speaker: (i+1)*sample_per_speaker]\n",
    "    X_train_per_speaker.append(np.concatenate(X_speaker).flatten())\n",
    "\n",
    "plt.boxplot(X_train_per_speaker)\n",
    "plt.xlabel(\"Speaker\")\n",
    "plt.ylabel(\"LPC (cepstra)\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Transduction (sequence-to-sequence model)\n",
    "\n",
    "As ReservoirPy Nodes are built to work on sequences, the simplest setup to solve this task is *sequence-to-sequence encoding*, also called *transduction*. A model is trained on encoding each vector of input sequence into a new vector in the output space. Thus, a sequence of audio yields a sequence of label, one label per timestep."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# repeat_target ensure that we obtain one label per timestep, and not one label per utterance.\n",
    "X_train, Y_train, X_test, Y_test = japanese_vowels(repeat_targets=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train a simple Echo State Network to solve this task:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from reservoirpy.nodes import Reservoir, Ridge, Input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "source = Input()\n",
    "reservoir = Reservoir(500, sr=0.9, lr=0.1)\n",
    "readout = Ridge(ridge=1e-6)\n",
    "\n",
    "model = [source >> reservoir, source] >> readout"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fit the model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y_pred = model.fit(X_train, Y_train, stateful=False, warmup=2).run(X_test, stateful=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Get the scores:\n",
    "\n",
    "There are 9 speakers, hence the output space is 9-dimensional. The speaker label is the index of the output neuron with maximum activation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:  91.173 %\n"
     ]
    }
   ],
   "source": [
    "Y_pred_class = [np.argmax(y_p, axis=1) for y_p in Y_pred]\n",
    "Y_test_class = [np.argmax(y_t, axis=1) for y_t in Y_test]\n",
    "\n",
    "score = accuracy_score(np.concatenate(Y_test_class, axis=0), np.concatenate(Y_pred_class, axis=0))\n",
    "\n",
    "print(\"Accuracy: \", f\"{score * 100:.3f} %\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Classification (sequence-to-vector model)\n",
    "\n",
    "We can create a more elaborated model where inference is performed only once on the whole input sequence. Indeed, we only need to assign one label to each input sequence. This new setup is known as a *sequence-to-vector* model, and this is usually the type of model we refer to when talking about classification of sequential patterns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, Y_train, X_test, Y_test = japanese_vowels()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "from reservoirpy.nodes import Reservoir, Ridge, Input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "source = Input()\n",
    "reservoir = Reservoir(500, sr=0.9, lr=0.1)\n",
    "readout = Ridge(ridge=1e-6)\n",
    "\n",
    "model = source >> reservoir >> readout"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We need to modify the training loop by hand a bit to perform this task:\n",
    "- first, we compute all reservoir states over the input sequence using the `reservoir.run` method.\n",
    "- then, we gather in a list only the last vector of the states sequence."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "states_train = []\n",
    "for x in X_train:\n",
    "    states = reservoir.run(x, reset=True)\n",
    "    states_train.append(states[-1, np.newaxis])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now train the readout only on the last state vectors. Here, `Y_train` is an array storing a single label for each utterance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Ridge-1': Ridge(ridge=1e-06, input_bias=True, in=500, out=9)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "readout.fit(states_train, Y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We also modify the inference code using the same method as above:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y_pred = []\n",
    "for x in X_test:\n",
    "    states = reservoir.run(x, reset=True)\n",
    "    y = readout.run(states[-1, np.newaxis])\n",
    "    Y_pred.append(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:  92.432 %\n"
     ]
    }
   ],
   "source": [
    "Y_pred_class = [np.argmax(y_p) for y_p in Y_pred]\n",
    "Y_test_class = [np.argmax(y_t) for y_t in Y_test]\n",
    "\n",
    "score = accuracy_score(Y_test_class, Y_pred_class)\n",
    "\n",
    "print(\"Accuracy: \", f\"{score * 100:.3f} %\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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",
   "version": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}