doc/source/methods/mmddrift.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[source](../api/alibi_detect.cd.mmd.rst)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Maximum Mean Discrepancy\n",
    "\n",
    "## Overview\n",
    "\n",
    "The [Maximum Mean Discrepancy (MMD)](http://jmlr.csail.mit.edu/papers/v13/gretton12a.html) detector is a kernel-based method for multivariate 2 sample testing. The MMD is a distance-based measure between 2 distributions *p* and *q* based on the mean embeddings $\\mu_{p}$ and $\\mu_{q}$ in a reproducing kernel Hilbert space $F$:\n",
    "\n",
    "\\begin{align}\n",
    "MMD(F, p, q) & = || \\mu_{p} - \\mu_{q} ||^2_{F} \\\\\n",
    "\\end{align}\n",
    "\n",
    "We can compute unbiased estimates of $MMD^2$ from the samples of the 2 distributions after applying the kernel trick. We use by default a [radial basis function kernel](https://en.wikipedia.org/wiki/Radial_basis_function_kernel), but users are free to pass their own kernel of preference to the detector. We obtain a $p$-value via a [permutation test](https://en.wikipedia.org/wiki/Resampling_(statistics)) on the values of $MMD^2$.\n",
    "\n",
    "For high-dimensional data, we typically want to reduce the dimensionality before computing the permutation test. Following suggestions in [Failing Loudly: An Empirical Study of Methods for Detecting Dataset Shift](https://arxiv.org/abs/1810.11953), we incorporate Untrained AutoEncoders (UAE), black-box shift detection using the classifier's softmax outputs ([BBSDs](https://arxiv.org/abs/1802.03916)) and [PCA](https://en.wikipedia.org/wiki/Principal_component_analysis) as out-of-the box preprocessing methods. Preprocessing methods which do not rely on the classifier will usually pick up drift in the input data, while BBSDs focuses on label shift.\n",
    "\n",
    "Detecting input data drift (covariate shift) $\\Delta p(x)$ for text data requires a custom preprocessing step. We can pick up changes in the semantics of the input by extracting (contextual) embeddings and detect drift on those. Strictly speaking we are not detecting $\\Delta p(x)$ anymore since the whole training procedure (objective function, training data etc) for the (pre)trained embeddings has an impact on the embeddings we extract. The library contains functionality to leverage pre-trained embeddings from [HuggingFace's transformer package](https://github.com/huggingface/transformers) but also allows you to easily use your own embeddings of choice. Both options are illustrated with examples in the [Text drift detection on IMDB movie reviews](../examples/cd_text_imdb.nblink) notebook."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Usage\n",
    "\n",
    "### Initialize\n",
    "\n",
    "\n",
    "Arguments:\n",
    "\n",
    "* `x_ref`: Data used as reference distribution.\n",
    "\n",
    "\n",
    "Keyword arguments:\n",
    "\n",
    "* `backend`: Both **TensorFlow** and **PyTorch** implementations of the MMD detector as well as various preprocessing steps are available. Specify the backend (*tensorflow* or *pytorch*). Defaults to *tensorflow*.\n",
    "\n",
    "* `p_val`: p-value used for significance of the permutation test.\n",
    "\n",
    "* `preprocess_x_ref`: Whether to already apply the (optional) preprocessing step to the reference data at initialization and store the preprocessed data. Dependent on the preprocessing step, this can reduce the computation time for the predict step significantly, especially when the reference dataset is large. Defaults to *True*. It is possible that it needs to be set to *False* if the preprocessing step requires statistics from both the reference and test data, such as the mean or standard deviation.\n",
    "\n",
    "* `update_x_ref`: Reference data can optionally be updated to the last N instances seen by the detector or via [reservoir sampling](https://en.wikipedia.org/wiki/Reservoir_sampling) with size N. For the former, the parameter equals *{'last': N}* while for reservoir sampling *{'reservoir_sampling': N}* is passed.\n",
    "\n",
    "* `preprocess_fn`: Function to preprocess the data before computing the data drift metrics. Typically a dimensionality reduction technique.\n",
    "\n",
    "* `kernel`: Kernel used when computing the MMD. Defaults to a Gaussian RBF kernel (`from alibi_detect.utils.pytorch import GaussianRBF` or `from alibi_detect.utils.tensorflow import GaussianRBF` dependent on the backend used).\n",
    "\n",
    "* `sigma`: Optional bandwidth for the kernel as a `np.ndarray`. We can also average over a number of different bandwidths, e.g. `np.array([.5, 1., 1.5])`.\n",
    "\n",
    "* `configure_kernel_from_x_ref`: If `sigma` is not specified, the detector can infer it via a heuristic and set `sigma` to the median pairwise distance between 2 samples. If `configure_kernel_from_x_ref` is *True*, we can already set `sigma` at initialization of the detector by inferring it from `x_ref`, speeding up the prediction step. If set to *False*, `sigma` is computed separately for each test batch at prediction time.\n",
    "\n",
    "* `n_permutations`: Number of permutations used in the permutation test.\n",
    "\n",
    "* `input_shape`: Optionally pass the shape of the input data.\n",
    "\n",
    "* `data_type`: can specify data type added to the metadata. E.g. *'tabular'* or *'image'*.\n",
    "\n",
    "\n",
    "Additional PyTorch keyword arguments:\n",
    "\n",
    "* `device`: *cuda* or *gpu* to use the GPU and *cpu* for the CPU. If the device is not specified, the detector will try to leverage the GPU if possible and otherwise fall back on CPU.\n",
    "\n",
    "\n",
    "Initialized drift detector example:\n",
    "\n",
    "\n",
    "```python\n",
    "from alibi_detect.cd import MMDDrift\n",
    "\n",
    "cd = MMDDrift(x_ref, backend='tensorflow', p_val=.05)\n",
    "```\n",
    "\n",
    "The same detector in PyTorch:\n",
    "\n",
    "```python\n",
    "cd = MMDDrift(x_ref, backend='pytorch', p_val=.05)\n",
    "```\n",
    "\n",
    "We can also easily add preprocessing functions for both frameworks. The following example uses a randomly initialized image encoder in PyTorch:\n",
    "\n",
    "```python\n",
    "from functools import partial\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from alibi_detect.cd.pytorch import preprocess_drift\n",
    "\n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "\n",
    "# define encoder\n",
    "encoder_net = nn.Sequential(\n",
    "    nn.Conv2d(3, 64, 4, stride=2, padding=0),\n",
    "    nn.ReLU(),\n",
    "    nn.Conv2d(64, 128, 4, stride=2, padding=0),\n",
    "    nn.ReLU(),\n",
    "    nn.Conv2d(128, 512, 4, stride=2, padding=0),\n",
    "    nn.ReLU(),\n",
    "    nn.Flatten(),\n",
    "    nn.Linear(2048, 32)\n",
    ").to(device).eval()\n",
    "\n",
    "# define preprocessing function\n",
    "preprocess_fn = partial(preprocess_drift, model=encoder_net, device=device, batch_size=512)\n",
    "\n",
    "cd = MMDDrift(x_ref, backend='pytorch', p_val=.05, preprocess_fn=preprocess_fn)\n",
    "```\n",
    "The same functionality is supported in TensorFlow and the main difference is that you would import from `alibi_detect.cd.tensorflow import preprocess_drift`. Other preprocessing steps such as the output of hidden layers of a model or extracted text embeddings using transformer models can be used in a similar way in both frameworks. TensorFlow example for the hidden layer output:\n",
    "\n",
    "```python\n",
    "from alibi_detect.cd.tensorflow import HiddenOutput, preprocess_drift\n",
    "\n",
    "model = # TensorFlow model; tf.keras.Model or tf.keras.Sequential\n",
    "preprocess_fn = partial(preprocess_drift, model=HiddenOutput(model, layer=-1), batch_size=128)\n",
    "\n",
    "cd = MMDDrift(x_ref, backend='tensorflow', p_val=.05, preprocess_fn=preprocess_fn)\n",
    "```\n",
    "\n",
    "Check out the [Drift detection on CIFAR10](../examples/cd_mmd_cifar10.nblink) example for more details.\n",
    "\n",
    "Alibi Detect also includes custom text preprocessing steps in both TensorFlow and PyTorch based on Huggingface's [transformers](https://github.com/huggingface/transformers) package:\n",
    "\n",
    "```python\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from transformers import AutoTokenizer\n",
    "from alibi_detect.cd.pytorch import preprocess_drift\n",
    "from alibi_detect.models.pytorch import TransformerEmbedding\n",
    "\n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "\n",
    "model_name = 'bert-base-cased'\n",
    "tokenizer = AutoTokenizer.from_pretrained(model_name)\n",
    "\n",
    "embedding_type = 'hidden_state'\n",
    "layers = [5, 6, 7]\n",
    "embed = TransformerEmbedding(model_name, embedding_type, layers)\n",
    "model = nn.Sequential(embed, nn.Linear(768, 256), nn.ReLU(), nn.Linear(256, enc_dim)).to(device).eval()\n",
    "preprocess_fn = partial(preprocess_drift, model=model, tokenizer=tokenizer, max_len=512, batch_size=32)\n",
    "\n",
    "# initialise drift detector\n",
    "cd = MMDDrift(x_ref, backend='pytorch', p_val=.05, preprocess_fn=preprocess_fn)\n",
    "```\n",
    "\n",
    "Again the same functionality is supported in TensorFlow but with `from alibi_detect.cd.tensorflow import preprocess_drift` and `from alibi_detect.models.tensorflow import TransformerEmbedding` imports. Check out the [Text drift detection on IMDB movie reviews](../examples/cd_text_imdb.nblink) example for more information."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Detect Drift\n",
    "\n",
    "We detect data drift by simply calling `predict` on a batch of instances `x`. We can return the p-value and the threshold of the permutation test by setting `return_p_val` to *True* and the maximum mean discrepancy metric and threshold by setting `return_distance` to *True*.\n",
    "\n",
    "The prediction takes the form of a dictionary with `meta` and `data` keys. `meta` contains the detector's metadata while `data` is also a dictionary which contains the actual predictions stored in the following keys:\n",
    "\n",
    "* `is_drift`: 1 if the sample tested has drifted from the reference data and 0 otherwise.\n",
    "\n",
    "* `p_val`: contains the p-value if `return_p_val` equals *True*.\n",
    "\n",
    "* `threshold`: p-value threshold if `return_p_val` equals *True*.\n",
    "\n",
    "* `distance`: MMD^2 metric between the reference data and the new batch if `return_distance` equals *True*.\n",
    "\n",
    "* `distance_threshold`: MMD^2 metric value from the permutation test which corresponds to the the p-value threshold.\n",
    "\n",
    "\n",
    "```python\n",
    "preds = cd.predict(X, return_p_val=True, return_distance=True)\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Saving and loading\n",
    "\n",
    "The drift detectors can be saved and loaded in the same way as other detectors:\n",
    "\n",
    "```python\n",
    "from alibi_detect.utils.saving import save_detector, load_detector\n",
    "\n",
    "filepath = 'my_path'\n",
    "save_detector(cd, filepath)\n",
    "cd = load_detector(filepath)\n",
    "```\n",
    "\n",
    "Currently on the **TensorFlow** backend is supported for `save_detector` and `load_detector`. Adding **PyTorch** support is a near term priority."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Examples\n",
    "\n",
    "### Image\n",
    "\n",
    "[Drift detection on CIFAR10](../examples/cd_mmd_cifar10.nblink)\n",
    "\n",
    "### Text\n",
    "\n",
    "[Text drift detection on IMDB movie reviews](../examples/cd_text_imdb.nblink)"
   ]
  }
 ],
 "metadata": {
  "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",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}