{ "cells": [ { "cell_type": "markdown", "metadata": { "slideshow": { "slide_type": "slide" } }, "source": [ "# A quick start to ReservoirPy\n", "\n", "Welcome to ReservoirPy!\n", "\n", "\n", "ReservoirPy is a Python tool designed to easily define, train and use Reservoir Computing (RC) architectures, such as Echo State Networks (ESNs).\n", "\n", "In the following tutorial, you will learn more about how to create an ESN with a few line of code, train it on some sequential data or timeseries and perform predictions. You will also learn more about ReservoirPy API, explained in more details in [this part of the documentation](https://reservoirpy.readthedocs.io/en/latest/user_guide/node.html).\n", "\n", "Once you have completed this tutorial, you can check the others and learn more about online learning rules, custom weight matrices, hyperparameters and how to choose them, and how to build complex architectures." ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import reservoirpy as rpy\n", "\n", "rpy.verbosity(0) # no need to be too verbose here\n", "rpy.set_seed(42) # make everything reproducible!" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## A note on data formats\n", "\n", "In ReservoirPy, **all data are stored in Numpy arrays**. It includes parameters of ReservoirPy Nodes and their input data. ReservoirPy uses only Numpy and Scipy for all computations.\n", "\n", "All timeseries, from input data to reservoir's activations, are formatted the same way: they must be Numpy arrays of shape $(timesteps, features)$. For instance, a timeseries composed of two variables samples over 100 timesteps would be stored in an array of shape $(100, 2)$. Single timesteps must also comply to this convention. A single timestep of the same 2-dimensional timeseries would be an array of shape $(1, 2)$. **Make sure to always comply with this formatting** otherwise you might obtain unexpected results, or raise an error.\n", "\n", "When training or running over several independent timeseries, for instance using several audio signals or several sine wave at different frequency, an accepted shape may be $(series, timesteps, features)$. In that case, input data may be an array where all series have the same length $timesteps$, or a list of arrays storing timeseries of different lengths." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Create your first Echo State Network\n", "\n", "Echo State Networks (ESNs) are one of the simplest recurrent neural network architecture RC has to offer. They are built using two components: a **reservoir** and a **readout**.\n", "\n", "The **reservoir** is a pool of artificial rate neurons randomly connected to their inputs and to themselves. Therefore, they are a genuine recurrent neural network similar to those which can be found in the field of machine learning in general. However, **none of the connections inside the reservoir are trained**. They are randomly initialized following several hyperparameters.\n", "\n", "The random high dimensional activation vector of the reservoir is then fed to the **readout**, a single layer of neurons in charge with decoding the reservoir activations in order to perform a task. The connections between the readout and the reservoir are the only **learned** connections in an ESN. In consequence, there is no need for complicated learning algorithm such as back-propagation of error to train them: a simple linear regression such as a regularized ridge regression can make it.\n", "\n", "Activations of the readout might also be fed back to the reservoir. This is called a **feedback** connection. This is an optional feature that can help tame the reservoir neurons activities.\n", "\n", "The figure below summarizes what an ESN looks like. Connections between neurons are stored as Numpy arrays or Scipy sparse matrices. These connection weights are here represented as colored arrows. $W_{in}$ and $W$ are the inputs-to-reservoir and recurrent connections. $W_{out}$ represents the trained output weights of the readout. $W_{fb}$ represents the additional connections feeding the readout outputs back to the reservoir.\n", "\n", "![An ESN.](../_static/user_guide/tutorials/esn.svg.png)\n", "\n", "In ReservoirPy, an ESN is the association of two components: a reservoir Node and a readout Node, connected together. In this example, the reservoir will be created using the `Reservoir` class, and the readout using a `Ridge` node, able to perform regularized linear regression on the reservoir's activations $x$.\n", "\n", "![An ESN in ReservoirPy](../_static/user_guide/tutorials/esn_nodes.svg.png)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Create a Reservoir\n", "\n", "We will first create a reservoir for our ESN, with 100 neurons. Reservoirs can be created using the `Reservoir` class.\n", "\n", "We change the value of two hyperparameters:\n", "- `lr`: the *leaking rate*, which controls the time constant of the neurons;\n", "- `sr`: the *spectral radius* of the recurrent connections in the reservoir. It controls the chaoticity of the reservoir dynamics.\n", "\n", "You can find all parameters and hyperparameters in [the `Reservoir` class documentation](https://reservoirpy.readthedocs.io/en/latest/api/generated/reservoirpy.nodes.Reservoir.html)." ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "from reservoirpy.nodes import Reservoir\n", "\n", "reservoir = Reservoir(100, lr=0.5, sr=0.9)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Initialize and run the reservoir" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This reservoir is empty for the moment: its parameters are not initialized yet. To start using the reservoir right now, you can run it on a timeseries, or use its `initialize()` method. Note that this method requires you to provide some samples of data, to infer the dimensions of some of the reservoir parameters, or to specify the input dimension of the reservoir at creation using the `input_dim` parameter.\n", "\n", "For now, let's run the reservoir on some data to visualize its activation.\n", "\n", "We will start by using an univariate timeseries as input to the reservoir, 100 discrete timesteps sampled from a sine wave for instance. This sine wave is stored in a Numpy array of shape $(100, 1)$." ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "import numpy as np\n", "import matplotlib.pyplot as plt\n", "\n", "X = np.sin(np.linspace(0, 6*np.pi, 100)).reshape(-1, 1)\n", "\n", "plt.figure(figsize=(10, 3))\n", "plt.title(\"A sine wave.\")\n", "plt.ylabel(\"$sin(t)$\")\n", "plt.xlabel(\"$t$\")\n", "plt.plot(X)\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Call on a single timestep\n", "\n", "All ReservoirPy nodes such as `Reservoir`, can run over single timesteps of data or complete timeseries.\n", "\n", "To run a node on a single timestep of data, use the call method:" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "New state vector shape: (1, 100)\n" ] } ], "source": [ "s = reservoir(X[0].reshape(1, -1))\n", "\n", "print(\"New state vector shape: \", s.shape)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This has triggered the reservoir neurons on a single timestep of timeseries $X$.\n", "\n", "Triggering a node on data not only returns the activations of that node, it also store this activation into the node internal state. This state can be accessed anytime using the `state()` method:" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [], "source": [ "s = reservoir.state()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This internal state is initialized to a null vector by default, and **will be updated every time you call the node on some data.** We can for instance perform successive calls to the reservoir to gather its activations of the whole timeseries:" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "states = np.empty((len(X), reservoir.output_dim))\n", "for i in range(len(X)):\n", " states[i] = reservoir(X[i].reshape(1, -1))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In the figure below, **we have plotted the activation of 20 neurons in the reservoir for every point of the timeseries**. Because the reservoir is a recurrent neural network, this state evolves in time following the evolution of the input timeseries." ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "plt.figure(figsize=(10, 3))\n", "plt.title(\"Activation of 20 reservoir neurons.\")\n", "plt.ylabel(\"$reservoir(sin(t))$\")\n", "plt.xlabel(\"$t$\")\n", "plt.plot(states[:, :20])\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Run over a whole timeseries\n", "\n", "Gathering the activations of a node over a timeseries can be done without using a for-loop with the `run()` method.\n", "This method takes arrays of shape $(timesteps, features)$ as input and returns an array of shape $(timesteps, states)$." ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [], "source": [ "states = reservoir.run(X)" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [], "source": [ "from reservoirpy.nodes import Reservoir, Ridge, FORCE, ESN" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "\n", "## Reset or modify reservoir state\n", "\n", "A node state can then be reset to a null vector to wash out its internal memory using the `reset()` method." ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [], "source": [ "reservoir = reservoir.reset()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "It is also possible to reset the state of a node when using the call or the `run()` method, setting the `reset` parameter to `True`:" ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [], "source": [ "states_from_null = reservoir.run(X, reset=True)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "States can also be fed to a node anytime using the `from_state` parameter of these methods. This allow for instance to reset a reservoir to a previous \"memory state\", or to a random one like in the code below:" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [], "source": [ "a_state_vector = np.random.uniform(-1, 1, size=(1, reservoir.output_dim))\n", "\n", "states_from_a_starting_state = reservoir.run(X, from_state=a_state_vector)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This operations can also be performed without erasing the node memory using the `with_state` context manager:" ] }, { "cell_type": "code", "execution_count": 13, "metadata": {}, "outputs": [], "source": [ "previous_states = reservoir.run(X)\n", "\n", "with reservoir.with_state(reset=True):\n", " states_from_null = reservoir.run(X)\n", "\n", "# as if the with_state never happened!\n", "states_from_previous = reservoir.run(X)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Everything happening inside the context manager will be erased when exiting it. The last state computed in `previous_states` is thus still available when computing `states_from_previous`, although the internal states was temporarily reset to 0 to compute `states_from_null`.\n", "\n", "These features allow to easily keep track of internal activations of nodes like reservoir, to modify them or to store them for later." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Create a readout\n", "\n", "Readouts works the same way as reservoirs. Internally, they are the same type of object, a *node*, and thus expose the same interface as described above: they have a call and a `run()` method, a `state()` (even if they are not necessarily a recurrent network), and so on.\n", "\n", "However, *offline readouts* such as readouts trained with linear regression are not really initialized at first call, like the reservoir. They need to be **fitted**: the connections going from the reservoir to the readout neurons need to be learned from data.\n", "\n", "Our first readout will be a simple [`Ridge` readout](https://reservoirpy.readthedocs.io/en/latest/api/generated/reservoirpy.nodes.Ridge.html#reservoirpy.nodes.Ridge), which can solve a task using linear regression.\n", "\n", "With set the `ridge` parameter of the readout to $10^{-7}$. This is the *regularization*, an hyperparameter that will help avoid overfitting.\n", "\n", "Note that we do not necessarily need to specify the number of output neurons in advance in a readout. It can be inferred at training time, and will be set to match the dimension of the teacher vectors." ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [], "source": [ "from reservoirpy.nodes import Ridge\n", "\n", "readout = Ridge(ridge=1e-7)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Define a training task\n", "\n", "Nodes like `Ridge` can be trained using their `fit()` method. This method takes as input two timeseries: a input timeseries and a target timeseries. During training, the readout will have to create a mapping from inputs to targets to solve a specific task: for instance, map a timestep $t$ of a timeseries to its timestep $t+1$. This would solve a one-timestep-ahead prediction task: from data at time $t$, predict data at time $t+1$.\n", "\n", "For the sake of example, let's solve this one-timestep-ahead prediction task. Given the sine wave defined above, can we predict its behavior in advance?\n", "\n", "To learn this task, we create a copy of the timeseries $X$ shifted by one timestep, to obtain a target timeseries." ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "X_train = X[:50]\n", "Y_train = X[1:51]\n", "\n", "plt.figure(figsize=(10, 3))\n", "plt.title(\"A sine wave and its future.\")\n", "plt.xlabel(\"$t$\")\n", "plt.plot(X_train, label=\"sin(t)\", color=\"blue\")\n", "plt.plot(Y_train, label=\"sin(t+1)\", color=\"red\")\n", "plt.legend()\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Given any point in the blue wave, we seek to predict the corresponding point in the red wave." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Train the readout\n", "\n", "While this is not the recommended way of using a readout node, let's train it as a standalone node.\n", "Our `Ridge` readout can be trained onĀ `train_states` and `Y_train`, where `train_states` are the activations of the reservoir triggered by `X_train`.\n", "\n", "First, we compute `train_states` using the `run()` method." ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [], "source": [ "train_states = reservoir.run(X_train, reset=True)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Then, we train the readout using `fit()`. We use the `warmup` parameter to set the amount of timesteps we want to discard at the beginning of `train_states`, when training. Indeed, as the internal state of the reservoir is recurrently updated bu start to a null value, it takes time before this state is representative of the input data." ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [], "source": [ "readout = readout.fit(train_states, Y_train, warmup=10)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In the plot below, we can see training was successful: our readout is trying to predict the sine next value!\n", "\n", "These results are not really impressive: a sine wave is easy to predict, and our readout still makes some mistakes. However, remember that we used only 50 timesteps of data to train that readout (40 actually, when discounting the 10 warmup steps). Reservoir and readouts together can obviously solve more complicated tasks, such as speech recognition or chaotic timeseries forecasting." ] }, { "cell_type": "code", "execution_count": 18, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "test_states = reservoir.run(X[50:])\n", "Y_pred = readout.run(test_states)\n", "\n", "plt.figure(figsize=(10, 3))\n", "plt.title(\"A sine wave and its future.\")\n", "plt.xlabel(\"$t$\")\n", "plt.plot(Y_pred, label=\"Predicted sin(t)\", color=\"blue\")\n", "plt.plot(X[51:], label=\"Real sin(t+1)\", color=\"red\")\n", "plt.legend()\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Create the ESN model\n", "\n", "Now that we have created a readout and a reservoir, and that we have seen how to run and train them, let's connect them to create an ESN. This section describes the \"canonical\" method for creating and training ESNs.\n", "\n", "In ReservoirPy, connecting nodes together leads to the creation of `Model` objects. Models are a specific type of node containing other nodes, and a description of all the connections between them. When run, a Model will run all the nodes it contains, and make the data follow their declared connections. When trained, it will also take care of training all its readouts.\n", "\n", "An ESN is a very simple type of Model, containing two nodes: a reservoir and a readout. When run, an ESN Model will first run the reservoir, feed the reservoir activations to the readout and run the readout on them.\n", "\n", "To declare connections between nodes and build a model, use the `>>` operator:" ] }, { "cell_type": "code", "execution_count": 19, "metadata": {}, "outputs": [], "source": [ "from reservoirpy.nodes import Reservoir, Ridge\n", "\n", "reservoir = Reservoir(100, lr=0.5, sr=0.9)\n", "readout = Ridge(ridge=1e-7)\n", "\n", "esn_model = reservoir >> readout" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Train the ESN" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Once the Model is created, it can be trained using its `fit()` method:" ] }, { "cell_type": "code", "execution_count": 20, "metadata": {}, "outputs": [], "source": [ "esn_model = esn_model.fit(X_train, Y_train, warmup=10)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "You can check that this method has initialized all nodes in the Model, and trained the `Ridge` readout." ] }, { "cell_type": "code", "execution_count": 21, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "True True True\n" ] } ], "source": [ "print(reservoir.is_initialized, readout.is_initialized, readout.fitted)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Run the ESN" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can then run the model on unseen next values on the timeseries to evaluate it, as shown below." ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "Y_pred = esn_model.run(X[50:])\n", "\n", "plt.figure(figsize=(10, 3))\n", "plt.title(\"A sine wave and its future.\")\n", "plt.xlabel(\"$t$\")\n", "plt.plot(Y_pred, label=\"Predicted sin(t+1)\", color=\"blue\")\n", "plt.plot(X[51:], label=\"Real sin(t+1)\", color=\"red\")\n", "plt.legend()\n", "plt.show()\n" ] } ], "metadata": { "celltoolbar": "Aucun(e)", "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": 4 }