diff --git a/examples/draft-point-est.ipynb b/examples/draft-point-est.ipynb new file mode 100644 index 000000000..2c54a8c20 --- /dev/null +++ b/examples/draft-point-est.ipynb @@ -0,0 +1,375 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "id": "045b991c-dd04-41b7-a345-366636238849", + "metadata": {}, + "source": [ + "# Draft: Point estimation by the example of quantiles\n", + "\n", + "This notebook is an adaptation of the [tutorial notebook](https://github.com/bayesflow-org/bayesflow/blob/dev/examples/Linear_Regression.ipynb) on linear regression by Paul Bürkner and Lars Kühmichel.\n", + "\n", + "We use the same simulator but do point estimation instead of learning the full posterior distribution. " + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "id": "066a9b3e-aedb-4e54-a3ab-e3c9f4e10c78", + "metadata": {}, + "outputs": [], + "source": [ + "%load_ext autoreload\n", + "%autoreload 2" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "id": "1d685bfa-e254-4bdd-9a58-45ea4158af54", + "metadata": { + "scrolled": true + }, + "outputs": [], + "source": [ + "import matplotlib.pyplot as plt\n", + "import numpy as np\n", + "import seaborn as sns\n", + "\n", + "np.set_printoptions(suppress=True)\n", + "\n", + "# ensure the backend is set\n", + "import os\n", + "if \"KERAS_BACKEND\" not in os.environ:\n", + " # set this to \"torch\", \"tensorflow\", or \"jax\"\n", + " os.environ[\"KERAS_BACKEND\"] = \"jax\"\n", + "\n", + "import keras\n", + "\n", + "# for BayesFlow devs: this ensures that the latest dev version can be found\n", + "import sys\n", + "sys.path.append('../')\n", + "\n", + "import bayesflow as bf" + ] + }, + { + "cell_type": "markdown", + "id": "c2e3d5b8-a003-474b-9fb9-9f92688c146b", + "metadata": {}, + "source": [ + "## Simulator for linear regression" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "id": "d88e5b0d-4ae2-4470-a6cd-90bf0e65211c", + "metadata": {}, + "outputs": [], + "source": [ + "def meta(batch_size):\n", + " # batch_size needs to be present but will be ignored here\n", + " # N: number of observation in a dataset\n", + " N = np.random.randint(5, 15)\n", + " return dict(N=N)\n", + "\n", + "def prior():\n", + " # beta: regression coefficients (intercept, slope)\n", + " beta = np.random.normal([2, 0], [3, 1])\n", + " # sigma: residual standard deviation\n", + " sigma = np.random.gamma(1, 1)\n", + " return dict(beta=beta, sigma=sigma)\n", + "\n", + "def likelihood(beta, sigma, N):\n", + " # x: predictor variable\n", + " x = np.random.normal(0, 1, size=N)\n", + " # y: response variable\n", + " y = np.random.normal(beta[0] + beta[1] * x, sigma, size=N)\n", + " return dict(y=y, x=x)" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "id": "e0035185-cba5-4fa2-a26b-5f4ffe4ddcf1", + "metadata": {}, + "outputs": [], + "source": [ + "simulator = bf.simulators.make_simulator([prior, likelihood], meta_fn=meta)" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "id": "8091dd5d-8fc8-4d8f-a57b-44371749a806", + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "5\n", + "(50, 2)\n", + "(50, 1)\n", + "(50, 5)\n", + "(50, 5)\n" + ] + } + ], + "source": [ + "# Generate a batch of three training samples\n", + "sample_data = simulator.sample(50)\n", + "print(sample_data[\"N\"])\n", + "print(sample_data[\"beta\"].shape)\n", + "print(sample_data[\"sigma\"].shape)\n", + "print(sample_data[\"x\"].shape)\n", + "print(sample_data[\"y\"].shape)" + ] + }, + { + "cell_type": "code", + "execution_count": 6, + "id": "42d0c12f-48ff-41d1-a496-f98bfd8564e5", + "metadata": {}, + "outputs": [], + "source": [ + "adapter = (\n", + " bf.Adapter()\n", + " .broadcast(\"N\", to=\"x\")\n", + " .as_set([\"x\", \"y\"])\n", + " .constrain(\"sigma\", lower=0)\n", + " .standardize(exclude=[\"N\"])\n", + " .apply(include=\"N\", forward=lambda n: np.sqrt(n), inverse=lambda n: n**2)\n", + " .concatenate([\"beta\", \"sigma\"], into=\"inference_variables\")\n", + " .concatenate([\"x\", \"y\"], into=\"summary_variables\")\n", + " .rename(\"N\", \"inference_conditions\")\n", + ")" + ] + }, + { + "cell_type": "markdown", + "id": "e3611a80-73c1-4a91-8f75-de29f1d7b3c5", + "metadata": {}, + "source": [ + "## Defining an optimization problem with a quantile scoring rule" + ] + }, + { + "cell_type": "code", + "execution_count": 7, + "id": "503fbeed-4744-421c-b48e-ce8af193e4c9", + "metadata": {}, + "outputs": [], + "source": [ + "inference_network = bf.networks.QuantileRegressor(quantile_levels=[0.1, 0.5, 0.9])\n", + "summary_network = bf.networks.DeepSet(depth=2)" + ] + }, + { + "cell_type": "code", + "execution_count": 8, + "id": "7a9314c1-7977-4ff2-b377-b487cf9c324d", + "metadata": {}, + "outputs": [], + "source": [ + "approximator = bf.ContinuousPointApproximator(\n", + " inference_network=inference_network,\n", + " summary_network=summary_network,\n", + " adapter=adapter,\n", + ")" + ] + }, + { + "cell_type": "code", + "execution_count": 9, + "id": "feb75eff-671d-4f98-a908-40c65100a41f", + "metadata": {}, + "outputs": [], + "source": [ + "epochs = 10\n", + "num_batches = 128\n", + "batch_size = 64\n", + "learning_rate = keras.optimizers.schedules.CosineDecay(5e-4, decay_steps=epochs*num_batches, alpha=1e-6)\n", + "optimizer = keras.optimizers.Adam(learning_rate=learning_rate, clipnorm=1.0)\n", + "approximator.compile(optimizer=optimizer)" + ] + }, + { + "cell_type": "markdown", + "id": "4c3f8b29-ebb0-4efa-a0fb-48a2dc798016", + "metadata": {}, + "source": [ + "## Training" + ] + }, + { + "cell_type": "code", + "execution_count": 10, + "id": "26a7b6ea-7d33-44fd-b6b6-89dccdfcb669", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "INFO:bayesflow:Building dataset from simulator instance of SequentialSimulator.\n", + "INFO:bayesflow:Using 12 data loading workers.\n", + "INFO:bayesflow:Building on a test batch.\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Epoch 1/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 124ms/step - loss: 0.1782 - loss/inference_loss: 0.1782 \n", + "Epoch 2/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 10ms/step - loss: 0.1141 - loss/inference_loss: 0.1141\n", + "Epoch 3/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 11ms/step - loss: 0.1013 - loss/inference_loss: 0.1013 \n", + "Epoch 4/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 10ms/step - loss: 0.0931 - loss/inference_loss: 0.0931 \n", + "Epoch 5/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 11ms/step - loss: 0.0903 - loss/inference_loss: 0.0903\n", + "Epoch 6/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 10ms/step - loss: 0.0881 - loss/inference_loss: 0.0881\n", + "Epoch 7/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 10ms/step - loss: 0.0857 - loss/inference_loss: 0.0857\n", + "Epoch 8/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 10ms/step - loss: 0.0854 - loss/inference_loss: 0.0854 \n", + "Epoch 9/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 12ms/step - loss: 0.0816 - loss/inference_loss: 0.0816\n", + "Epoch 10/10\n", + "\u001b[1m128/128\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 12ms/step - loss: 0.0813 - loss/inference_loss: 0.0813\n" + ] + } + ], + "source": [ + "history = approximator.fit(\n", + " epochs=epochs,\n", + " num_batches=num_batches,\n", + " batch_size=batch_size,\n", + " simulator=simulator,\n", + ")" + ] + }, + { + "cell_type": "code", + "execution_count": 11, + "id": "0c9c3d05-9f48-4fd0-9119-6aabad750752", + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "# Visualize losses\n", + "f = bf.diagnostics.plots.loss(\n", + " train_losses=np.array(history.history[\"loss\"]).astype(np.float32)\n", + ")" + ] + }, + { + "cell_type": "markdown", + "id": "cd29ac9f-31fc-4a11-8943-7d9d47d0007d", + "metadata": {}, + "source": [ + "## Preliminary plotting of recovery" + ] + }, + { + "cell_type": "code", + "execution_count": 12, + "id": "c311edee-e9fc-448f-99d4-7261e3b4397a", + "metadata": {}, + "outputs": [], + "source": [ + "pred_quantiles = approximator.estimate(conditions={k:v for k,v in sample_data.items() if k in [\"N\", \"x\", \"y\"]})" + ] + }, + { + "cell_type": "code", + "execution_count": 13, + "id": "70405a2f-8978-4a7b-8431-c6fd7f70e3af", + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "# basic plots just to take a look at recovery\n", + "inference_variables = [(\"beta\", 0), (\"beta\",1), (\"sigma\",0)]\n", + "fig, axes = plt.subplots(1,len(inference_variables), figsize=(8,5))\n", + "\n", + "for i,ax in enumerate(axes):\n", + " ax.set_aspect('equal')\n", + " \n", + " inference_variable, var_idx = inference_variables[i]\n", + " \n", + " for j in range(sample_data[inference_variable].shape[0]):\n", + " num_quantiles = pred_quantiles[inference_variable].shape[1]\n", + " x = [sample_data[inference_variable][j,var_idx]]*num_quantiles\n", + " y = pred_quantiles[inference_variable][j,:,var_idx]\n", + " ax.plot(x, y, marker=\"_\")\n", + " \n", + " low_corner, high_corner = np.min(pred_quantiles[inference_variable][:,var_idx]), np.max(pred_quantiles[inference_variable][:,:,var_idx])\n", + " ax.plot([low_corner, high_corner], [low_corner, high_corner], color=\"black\", linewidth=0.8, linestyle=\"--\")\n", + " title = f\"$\\{inference_variable}_{var_idx}$\" \n", + " ax.set_title(title)" + ] + }, + { + "cell_type": "markdown", + "id": "a75703d2-ec29-4d31-b207-b835f42b3822", + "metadata": {}, + "source": [ + "Above the horizontal bars of the same color correspond to the 0.1, 0.5 and 0.9 quantiles respectively." + ] + } + ], + "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": "3.11.11" + }, + "widgets": { + "application/vnd.jupyter.widget-state+json": { + "state": {}, + "version_major": 2, + "version_minor": 0 + } + } + }, + "nbformat": 4, + "nbformat_minor": 5 +}