rvga-logreg.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "bf07e398-fd77-41e7-ab21-825b6a148ed0",
   "metadata": {},
   "source": [
    "# R-VGA for logistic regression\n",
    "\n",
    "$$\n",
    "\\begin{aligned}\n",
    "    \\mu_t &= \\mu_{t-1} + P_{t-1}\\left( y_t - \\mathbb{E}\\left[\\sigma(\\theta^\\intercal{\\bf x}_t\\right] \\right){\\bf x}_t\\\\\n",
    "    P_{t}^{-1} &= P_{t-1}^{-1} - \\mathbb{E}\\left[\\sigma'\\left(\\theta^\\intercal {\\bf x}_t\\right)\\right] x_tx_t^\\intercal\n",
    "\\end{aligned}\n",
    "$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "50da074a-2085-4763-9df1-f0ab345635cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "import jax\n",
    "import chex\n",
    "import distrax\n",
    "import jax.numpy as jnp\n",
    "import matplotlib.pyplot as plt\n",
    "from functools import partial"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "50d26f9e-1314-4f46-b960-00605976e8a4",
   "metadata": {},
   "outputs": [],
   "source": [
    "%config InlineBackend.figure_format = \"retina\"\n",
    "plt.rcParams[\"axes.spines.right\"] = False\n",
    "plt.rcParams[\"axes.spines.top\"] = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 316,
   "id": "c500ae10-9b5f-49be-9356-8d4115b53a44",
   "metadata": {},
   "outputs": [],
   "source": [
    "key = jax.random.PRNGKey(31)\n",
    "key_params, key_train, key_test = jax.random.split(key, 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 329,
   "id": "7383a59d-603f-4859-8f42-6321c9ffb5c4",
   "metadata": {},
   "outputs": [],
   "source": [
    "@chex.dataclass\n",
    "class Params:\n",
    "    mean: jnp.DeviceArray\n",
    "    prec: jnp.DeviceArray\n",
    "\n",
    "def sample_true_params(key, d, c=1.0):\n",
    "    key_mean, key_cov = jax.random.split(key)\n",
    "    \n",
    "    mean1 = jax.random.uniform(key_mean, minval=-3, maxval=3)\n",
    "    mean2 = -mean1\n",
    "    \n",
    "    M = jax.random.orthogonal(key_cov, d)\n",
    "    cov_diag = 1 / jnp.arange(1, d + 1) ** c\n",
    "    cov = jnp.einsum(\"ji,j,jk->ik\", M, cov_diag, M, precision=\"highest\")\n",
    "    \n",
    "    return mean1, mean2, cov\n",
    "\n",
    "def sample_dataset(key, mu1, mu2, cov, n_samps):\n",
    "    key_data, key_choice = jax.random.split(key, 2)\n",
    "    n_samp = n_train // 2\n",
    "    \n",
    "    X = jax.random.multivariate_normal(key_data, mean=jnp.zeros(d), cov=cov, shape=(n_train,))\n",
    "    choice = jax.random.choice(key_choice, n_train, replace=False, shape=(n_train,))\n",
    "\n",
    "    X = X.at[:n_samp].set(X[:n_samp] + mu1)\n",
    "    X = X.at[n_samp:].set(X[n_samp:] + mu2)\n",
    "    \n",
    "    y = jnp.zeros(len(X))\n",
    "    y = y.at[:n_samp].set(1.0)\n",
    "\n",
    "    X = X[choice]\n",
    "    y = y[choice]\n",
    "    \n",
    "    return X, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 394,
   "id": "40bda387-d859-497f-9125-2b78356b3ee5",
   "metadata": {},
   "outputs": [],
   "source": [
    "def sample_params(key, mu, cov):\n",
    "    params = distrax.MultivariateNormalFullCovariance(mu, cov).sample(seed=key, sample_shape=(n_samples))\n",
    "    return params\n",
    "\n",
    "\n",
    "def sample_pred(params, x):\n",
    "    \"\"\"\n",
    "    Predict the linear term w^T x—\n",
    "    the input to the logistic\n",
    "    \"\"\"\n",
    "    samples = jnp.einsum(\"sm,...m->...s\", params, x)\n",
    "    return samples\n",
    "    \n",
    "\n",
    "@jax.jit\n",
    "def expected_sigmoid(key, x, mu, cov):\n",
    "    params = sample_params(key, mu, cov)\n",
    "    x_samps = sample_pred(params, x)\n",
    "    y_out = jax.nn.sigmoid(x_samps)\n",
    "    \n",
    "    return y_out.mean(axis=-1)\n",
    "\n",
    "@jax.jit\n",
    "def expected_dsigmoid(key, x, mu, cov):\n",
    "    params = sample_params(key, mu, cov)\n",
    "    x_samps = sample_pred(params, x)\n",
    "    \n",
    "    y_out = jax.nn.sigmoid(x_samps)\n",
    "    y_out = y_out * (1 - y_out)\n",
    "    \n",
    "    return y_out.mean(axis=-1)\n",
    "\n",
    "\n",
    "@partial(jax.vmap, in_axes=(None, 0))\n",
    "def lossfn(y, phat):\n",
    "    logl = y * jnp.log(phat) + (1 - y) * jnp.log(1 - phat)\n",
    "    return -logl.mean(axis=-1)\n",
    "\n",
    "def _rva_inner(i, params, key, x, y):\n",
    "    key_i = jax.random.fold_in(key, i)\n",
    "    key_s, key_ds = jax.random.split(key_i, 2)\n",
    "    \n",
    "    mean = params.mean\n",
    "    prec = params.prec\n",
    "    cov = jnp.linalg.inv(prec)\n",
    "    \n",
    "    e_sigmoid = expected_sigmoid(key_s, x, mean, cov)\n",
    "    e_dsigmoid = expected_dsigmoid(key_ds, x, mean, cov)\n",
    "    \n",
    "    mean = mean + cov @ x * (y - e_sigmoid)\n",
    "    prec = prec + jnp.einsum(\"i,j->ij\", x, x) * e_dsigmoid\n",
    "    \n",
    "    params = params.replace(\n",
    "        mean=mean,\n",
    "        prec=prec\n",
    "    )\n",
    "    \n",
    "    return params\n",
    "\n",
    "@partial(jax.jit, static_argnames=(\"n_iter\",))\n",
    "def rvga_step(params, data, n_iter):\n",
    "    key, x, y = data\n",
    "    \n",
    "    _inner = partial(_rva_inner, x=x, y=y, key=key)\n",
    "    params = jax.lax.fori_loop(0, n_iter, _inner, params)\n",
    "    return params, params\n",
    "\n",
    "\n",
    "@jax.jit\n",
    "def ekf_step(params, data):\n",
    "    \"\"\"\n",
    "    The extended Kalman filter (EKF) for logistic regressino\n",
    "    See (81) in R-VGA paper.\n",
    "    \"\"\"\n",
    "    x, y = data\n",
    "    mean = params.mean\n",
    "    cov = jnp.linalg.inv(params.prec)\n",
    "    \n",
    "    yhat = mean @ x\n",
    "    phat = jax.nn.sigmoid(yhat)\n",
    "    dphat = phat * (1 - phat)\n",
    "    \n",
    "    # Denominator term\n",
    "    cov_den = 1 / dphat + jnp.einsum(\"i,ij,j->\", x, cov, x)\n",
    "    # Numerator term\n",
    "    cov_num = jnp.einsum(\"ij,j,k,kl->il\", cov, x, x, cov)\n",
    "    \n",
    "    cov = cov - cov_num / cov_den\n",
    "    mean = mean + cov @ x * (y - phat)\n",
    "    \n",
    "    prec = jnp.linalg.inv(cov)\n",
    "    params = params.replace(\n",
    "        mean=mean,\n",
    "        prec=prec\n",
    "    )\n",
    "    \n",
    "    return params, params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 421,
   "id": "ece8ee43-fad3-4300-86c5-dd36dbf7b958",
   "metadata": {},
   "outputs": [],
   "source": [
    "d = 3\n",
    "mu1, mu2, cov = sample_true_params(key_params, d, c=2.0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 422,
   "id": "1e964d71-7289-4e12-90b2-b9b05445e9a0",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_train = 200\n",
    "X, y = sample_dataset(key_train, mu1, mu2, cov, n_train)\n",
    "\n",
    "n_test = 100\n",
    "X_test, y_test = sample_dataset(key_test, mu1, mu2, cov, n_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 423,
   "id": "035da77b-3e01-4d40-97df-6aa47420947b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# plt.scatter(*X.T, c=y, cmap=\"Dark2\")\n",
    "# plt.grid(alpha=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 424,
   "id": "bbd03c9f-4272-47db-80e0-0f282bafbb62",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_samples = 100\n",
    "sigma2 = 1.0\n",
    "mu0 = jnp.zeros(d)\n",
    "prec0 = jnp.eye(d) / sigma2\n",
    "params0 = Params(mean=mu0, prec=prec0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 425,
   "id": "6ca46a3e-9db3-4900-98e4-18af9768ae8c",
   "metadata": {},
   "outputs": [],
   "source": [
    "keys = jax.random.split(key, n_obs)\n",
    "xs = (keys, X, y)\n",
    "\n",
    "xs_init = jax.tree_map(lambda x: x[0], xs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 426,
   "id": "d1dd91e1-337a-4204-979b-4de0b408b1cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_iter = 2\n",
    "keys = jax.random.split(key, n_obs)\n",
    "xs = (keys, X, y)\n",
    "\n",
    "bel, hist = jax.lax.scan(\n",
    "    partial(rvga_step, n_iter=1),\n",
    "    params0,\n",
    "    xs\n",
    ")\n",
    "\n",
    "xs = (X, y)\n",
    "bel_ekf, hist_ekf = jax.lax.scan(\n",
    "    ekf_step,\n",
    "    params0,\n",
    "    xs\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 427,
   "id": "b3f68d02-5f70-4438-be04-ae94e89d06ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "hist_cov = jax.vmap(jnp.linalg.inv)(hist.prec)\n",
    "\n",
    "hist_cov_ekf = jax.vmap(jnp.linalg.inv)(hist_ekf.prec)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 428,
   "id": "e1ea2463-3306-40d1-9143-8110b6c3e8e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "yhat = jax.vmap(expected_sigmoid, (None, None, 0, 0))(key, X_test, hist.mean, hist_cov)\n",
    "yhat_ekf = jax.vmap(expected_sigmoid, (None, None, 0, 0))(key, X_test, hist_ekf.mean, hist_cov_ekf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 429,
   "id": "090228b1-b206-4625-a329-cccae0109501",
   "metadata": {},
   "outputs": [],
   "source": [
    "hist_loss = lossfn(y_test, yhat)\n",
    "hist_acc = (yhat.round() == y_test[None, :]).mean(axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 430,
   "id": "080384dd-e0b4-4161-aa67-e384c7b2f2b6",
   "metadata": {},
   "outputs": [],
   "source": [
    "hist_loss_ekf = lossfn(y_test, yhat_ekf)\n",
    "hist_acc_ekf = (yhat_ekf.round() == y_test[None, :]).mean(axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 431,
   "id": "1bb1c5b0-c710-452f-b0e5-19a1bf6322d7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Accuracy')"
      ]
     },
     "execution_count": 431,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x300 with 2 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 296,
       "width": 826
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(1, 2, figsize=(10, 3)) \n",
    "\n",
    "ax[0].plot(hist_loss)\n",
    "ax[0].plot(hist_loss_ekf)\n",
    "\n",
    "ax[0].grid(alpha=0.2)\n",
    "ax[0].set_title(\"Loss\")\n",
    "\n",
    "ax[1].plot(hist_acc)\n",
    "ax[1].plot(hist_acc_ekf)\n",
    "ax[1].grid(alpha=0.2)\n",
    "ax[1].set_title(\"Accuracy\")"
   ]
  }
 ],
 "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": "3.10.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}