sklearn/doc/modules/kernel_approximation.rst

312 lines
14 KiB
ReStructuredText

.. _kernel_approximation:
Kernel Approximation
====================
This submodule contains functions that approximate the feature mappings that
correspond to certain kernels, as they are used for example in support vector
machines (see :ref:`svm`).
The following feature functions perform non-linear transformations of the
input, which can serve as a basis for linear classification or other
algorithms.
.. currentmodule:: sklearn.linear_model
The advantage of using approximate explicit feature maps compared to the
`kernel trick <https://en.wikipedia.org/wiki/Kernel_trick>`_,
which makes use of feature maps implicitly, is that explicit mappings
can be better suited for online learning and can significantly reduce the cost
of learning with very large datasets.
Standard kernelized SVMs do not scale well to large datasets, but using an
approximate kernel map it is possible to use much more efficient linear SVMs.
In particular, the combination of kernel map approximations with
:class:`SGDClassifier` can make non-linear learning on large datasets possible.
Since there has not been much empirical work using approximate embeddings, it
is advisable to compare results against exact kernel methods when possible.
.. seealso::
:ref:`polynomial_regression` for an exact polynomial transformation.
.. currentmodule:: sklearn.kernel_approximation
.. _nystroem_kernel_approx:
Nystroem Method for Kernel Approximation
----------------------------------------
The Nystroem method, as implemented in :class:`Nystroem` is a general method for
reduced rank approximations of kernels. It achieves this by subsampling without
replacement rows/columns of the data on which the kernel is evaluated. While the
computational complexity of the exact method is
:math:`\mathcal{O}(n^3_{\text{samples}})`, the complexity of the approximation
is :math:`\mathcal{O}(n^2_{\text{components}} \cdot n_{\text{samples}})`, where
one can set :math:`n_{\text{components}} \ll n_{\text{samples}}` without a
significative decrease in performance [WS2001]_.
We can construct the eigendecomposition of the kernel matrix :math:`K`, based
on the features of the data, and then split it into sampled and unsampled data
points.
.. math::
K = U \Lambda U^T
= \begin{bmatrix} U_1 \\ U_2\end{bmatrix} \Lambda \begin{bmatrix} U_1 \\ U_2 \end{bmatrix}^T
= \begin{bmatrix} U_1 \Lambda U_1^T & U_1 \Lambda U_2^T \\ U_2 \Lambda U_1^T & U_2 \Lambda U_2^T \end{bmatrix}
\equiv \begin{bmatrix} K_{11} & K_{12} \\ K_{21} & K_{22} \end{bmatrix}
where:
* :math:`U` is orthonormal
* :math:`\Lambda` is diagonal matrix of eigenvalues
* :math:`U_1` is orthonormal matrix of samples that were chosen
* :math:`U_2` is orthonormal matrix of samples that were not chosen
Given that :math:`U_1 \Lambda U_1^T` can be obtained by orthonormalization of
the matrix :math:`K_{11}`, and :math:`U_2 \Lambda U_1^T` can be evaluated (as
well as its transpose), the only remaining term to elucidate is
:math:`U_2 \Lambda U_2^T`. To do this we can express it in terms of the already
evaluated matrices:
.. math::
\begin{align} U_2 \Lambda U_2^T &= \left(K_{21} U_1 \Lambda^{-1}\right) \Lambda \left(K_{21} U_1 \Lambda^{-1}\right)^T
\\&= K_{21} U_1 (\Lambda^{-1} \Lambda) \Lambda^{-1} U_1^T K_{21}^T
\\&= K_{21} U_1 \Lambda^{-1} U_1^T K_{21}^T
\\&= K_{21} K_{11}^{-1} K_{21}^T
\\&= \left( K_{21} K_{11}^{-\frac12} \right) \left( K_{21} K_{11}^{-\frac12} \right)^T
.\end{align}
During ``fit``, the class :class:`Nystroem` evaluates the basis :math:`U_1`, and
computes the normalization constant, :math:`K_{11}^{-\frac12}`. Later, during
``transform``, the kernel matrix is determined between the basis (given by the
`components_` attribute) and the new data points, ``X``. This matrix is then
multiplied by the ``normalization_`` matrix for the final result.
By default :class:`Nystroem` uses the ``rbf`` kernel, but it can use any kernel
function or a precomputed kernel matrix. The number of samples used - which is
also the dimensionality of the features computed - is given by the parameter
``n_components``.
.. rubric:: Examples
* See the example entitled
:ref:`sphx_glr_auto_examples_applications_plot_cyclical_feature_engineering.py`,
that shows an efficient machine learning pipeline that uses a
:class:`Nystroem` kernel.
.. _rbf_kernel_approx:
Radial Basis Function Kernel
----------------------------
The :class:`RBFSampler` constructs an approximate mapping for the radial basis
function kernel, also known as *Random Kitchen Sinks* [RR2007]_. This
transformation can be used to explicitly model a kernel map, prior to applying
a linear algorithm, for example a linear SVM::
>>> from sklearn.kernel_approximation import RBFSampler
>>> from sklearn.linear_model import SGDClassifier
>>> X = [[0, 0], [1, 1], [1, 0], [0, 1]]
>>> y = [0, 0, 1, 1]
>>> rbf_feature = RBFSampler(gamma=1, random_state=1)
>>> X_features = rbf_feature.fit_transform(X)
>>> clf = SGDClassifier(max_iter=5)
>>> clf.fit(X_features, y)
SGDClassifier(max_iter=5)
>>> clf.score(X_features, y)
1.0
The mapping relies on a Monte Carlo approximation to the
kernel values. The ``fit`` function performs the Monte Carlo sampling, whereas
the ``transform`` method performs the mapping of the data. Because of the
inherent randomness of the process, results may vary between different calls to
the ``fit`` function.
The ``fit`` function takes two arguments:
``n_components``, which is the target dimensionality of the feature transform,
and ``gamma``, the parameter of the RBF-kernel. A higher ``n_components`` will
result in a better approximation of the kernel and will yield results more
similar to those produced by a kernel SVM. Note that "fitting" the feature
function does not actually depend on the data given to the ``fit`` function.
Only the dimensionality of the data is used.
Details on the method can be found in [RR2007]_.
For a given value of ``n_components`` :class:`RBFSampler` is often less accurate
as :class:`Nystroem`. :class:`RBFSampler` is cheaper to compute, though, making
use of larger feature spaces more efficient.
.. figure:: ../auto_examples/miscellaneous/images/sphx_glr_plot_kernel_approximation_002.png
:target: ../auto_examples/miscellaneous/plot_kernel_approximation.html
:scale: 50%
:align: center
Comparing an exact RBF kernel (left) with the approximation (right)
.. rubric:: Examples
* :ref:`sphx_glr_auto_examples_miscellaneous_plot_kernel_approximation.py`
.. _additive_chi_kernel_approx:
Additive Chi Squared Kernel
---------------------------
The additive chi squared kernel is a kernel on histograms, often used in computer vision.
The additive chi squared kernel as used here is given by
.. math::
k(x, y) = \sum_i \frac{2x_iy_i}{x_i+y_i}
This is not exactly the same as :func:`sklearn.metrics.pairwise.additive_chi2_kernel`.
The authors of [VZ2010]_ prefer the version above as it is always positive
definite.
Since the kernel is additive, it is possible to treat all components
:math:`x_i` separately for embedding. This makes it possible to sample
the Fourier transform in regular intervals, instead of approximating
using Monte Carlo sampling.
The class :class:`AdditiveChi2Sampler` implements this component wise
deterministic sampling. Each component is sampled :math:`n` times, yielding
:math:`2n+1` dimensions per input dimension (the multiple of two stems
from the real and complex part of the Fourier transform).
In the literature, :math:`n` is usually chosen to be 1 or 2, transforming
the dataset to size ``n_samples * 5 * n_features`` (in the case of :math:`n=2`).
The approximate feature map provided by :class:`AdditiveChi2Sampler` can be combined
with the approximate feature map provided by :class:`RBFSampler` to yield an approximate
feature map for the exponentiated chi squared kernel.
See the [VZ2010]_ for details and [VVZ2010]_ for combination with the :class:`RBFSampler`.
.. _skewed_chi_kernel_approx:
Skewed Chi Squared Kernel
-------------------------
The skewed chi squared kernel is given by:
.. math::
k(x,y) = \prod_i \frac{2\sqrt{x_i+c}\sqrt{y_i+c}}{x_i + y_i + 2c}
It has properties that are similar to the exponentiated chi squared kernel
often used in computer vision, but allows for a simple Monte Carlo
approximation of the feature map.
The usage of the :class:`SkewedChi2Sampler` is the same as the usage described
above for the :class:`RBFSampler`. The only difference is in the free
parameter, that is called :math:`c`.
For a motivation for this mapping and the mathematical details see [LS2010]_.
.. _polynomial_kernel_approx:
Polynomial Kernel Approximation via Tensor Sketch
-------------------------------------------------
The :ref:`polynomial kernel <polynomial_kernel>` is a popular type of kernel
function given by:
.. math::
k(x, y) = (\gamma x^\top y +c_0)^d
where:
* ``x``, ``y`` are the input vectors
* ``d`` is the kernel degree
Intuitively, the feature space of the polynomial kernel of degree `d`
consists of all possible degree-`d` products among input features, which enables
learning algorithms using this kernel to account for interactions between features.
The TensorSketch [PP2013]_ method, as implemented in :class:`PolynomialCountSketch`, is a
scalable, input data independent method for polynomial kernel approximation.
It is based on the concept of Count sketch [WIKICS]_ [CCF2002]_ , a dimensionality
reduction technique similar to feature hashing, which instead uses several
independent hash functions. TensorSketch obtains a Count Sketch of the outer product
of two vectors (or a vector with itself), which can be used as an approximation of the
polynomial kernel feature space. In particular, instead of explicitly computing
the outer product, TensorSketch computes the Count Sketch of the vectors and then
uses polynomial multiplication via the Fast Fourier Transform to compute the
Count Sketch of their outer product.
Conveniently, the training phase of TensorSketch simply consists of initializing
some random variables. It is thus independent of the input data, i.e. it only
depends on the number of input features, but not the data values.
In addition, this method can transform samples in
:math:`\mathcal{O}(n_{\text{samples}}(n_{\text{features}} + n_{\text{components}} \log(n_{\text{components}})))`
time, where :math:`n_{\text{components}}` is the desired output dimension,
determined by ``n_components``.
.. rubric:: Examples
* :ref:`sphx_glr_auto_examples_kernel_approximation_plot_scalable_poly_kernels.py`
.. _tensor_sketch_kernel_approx:
Mathematical Details
--------------------
Kernel methods like support vector machines or kernelized
PCA rely on a property of reproducing kernel Hilbert spaces.
For any positive definite kernel function :math:`k` (a so called Mercer kernel),
it is guaranteed that there exists a mapping :math:`\phi`
into a Hilbert space :math:`\mathcal{H}`, such that
.. math::
k(x,y) = \langle \phi(x), \phi(y) \rangle
Where :math:`\langle \cdot, \cdot \rangle` denotes the inner product in the
Hilbert space.
If an algorithm, such as a linear support vector machine or PCA,
relies only on the scalar product of data points :math:`x_i`, one may use
the value of :math:`k(x_i, x_j)`, which corresponds to applying the algorithm
to the mapped data points :math:`\phi(x_i)`.
The advantage of using :math:`k` is that the mapping :math:`\phi` never has
to be calculated explicitly, allowing for arbitrary large
features (even infinite).
One drawback of kernel methods is, that it might be necessary
to store many kernel values :math:`k(x_i, x_j)` during optimization.
If a kernelized classifier is applied to new data :math:`y_j`,
:math:`k(x_i, y_j)` needs to be computed to make predictions,
possibly for many different :math:`x_i` in the training set.
The classes in this submodule allow to approximate the embedding
:math:`\phi`, thereby working explicitly with the representations
:math:`\phi(x_i)`, which obviates the need to apply the kernel
or store training examples.
.. rubric:: References
.. [WS2001] `"Using the Nyström method to speed up kernel machines"
<https://papers.nips.cc/paper_files/paper/2000/hash/19de10adbaa1b2ee13f77f679fa1483a-Abstract.html>`_
Williams, C.K.I.; Seeger, M. - 2001.
.. [RR2007] `"Random features for large-scale kernel machines"
<https://papers.nips.cc/paper/2007/hash/013a006f03dbc5392effeb8f18fda755-Abstract.html>`_
Rahimi, A. and Recht, B. - Advances in neural information processing 2007,
.. [LS2010] `"Random Fourier approximations for skewed multiplicative histogram kernels"
<https://www.researchgate.net/publication/221114584_Random_Fourier_Approximations_for_Skewed_Multiplicative_Histogram_Kernels>`_
Li, F., Ionescu, C., and Sminchisescu, C.
- Pattern Recognition, DAGM 2010, Lecture Notes in Computer Science.
.. [VZ2010] `"Efficient additive kernels via explicit feature maps"
<https://www.robots.ox.ac.uk/~vgg/publications/2011/Vedaldi11/vedaldi11.pdf>`_
Vedaldi, A. and Zisserman, A. - Computer Vision and Pattern Recognition 2010
.. [VVZ2010] `"Generalized RBF feature maps for Efficient Detection"
<https://www.robots.ox.ac.uk/~vgg/publications/2010/Sreekanth10/sreekanth10.pdf>`_
Vempati, S. and Vedaldi, A. and Zisserman, A. and Jawahar, CV - 2010
.. [PP2013] :doi:`"Fast and scalable polynomial kernels via explicit feature maps"
<10.1145/2487575.2487591>`
Pham, N., & Pagh, R. - 2013
.. [CCF2002] `"Finding frequent items in data streams"
<https://www.cs.princeton.edu/courses/archive/spring04/cos598B/bib/CharikarCF.pdf>`_
Charikar, M., Chen, K., & Farach-Colton - 2002
.. [WIKICS] `"Wikipedia: Count sketch"
<https://en.wikipedia.org/wiki/Count_sketch>`_