Skip to content

Instantly share code, notes, and snippets.

Avatar

Mathieu Blondel mblondel

View GitHub Profile
@mblondel
mblondel / kmeans.py
Last active Jan 19, 2021
Fuzzy K-means and K-medians
View kmeans.py
# Copyright Mathieu Blondel December 2011
# License: BSD 3 clause
import numpy as np
import pylab as pl
from sklearn.base import BaseEstimator
from sklearn.utils import check_random_state
from sklearn.cluster import MiniBatchKMeans
from sklearn.cluster import KMeans as KMeansGood
View multiclass_svm.py
"""
Multiclass SVMs (Crammer-Singer formulation).
A pure Python re-implementation of:
Large-scale Multiclass Support Vector Machine Training via Euclidean Projection onto the Simplex.
Mathieu Blondel, Akinori Fujino, and Naonori Ueda.
ICPR 2014.
http://www.mblondel.org/publications/mblondel-icpr2014.pdf
"""
View kernel_sgd.py
# Mathieu Blondel, May 2012
# License: BSD 3 clause
import numpy as np
def euclidean_distances(X, Y=None, Y_norm_squared=None, squared=False):
XX = np.sum(X * X, axis=1)[:, np.newaxis]
YY = np.sum(Y ** 2, axis=1)[np.newaxis, :]
distances = np.dot(X, Y.T)
distances *= -2
@mblondel
mblondel / projection_simplex_vectorized.py
Last active Oct 15, 2020
Vectorized projection onto the simplex
View projection_simplex_vectorized.py
# Author: Mathieu Blondel
# License: BSD 3 clause
import numpy as np
def projection_simplex(V, z=1, axis=None):
"""
Projection of x onto the simplex, scaled by z:
P(x; z) = argmin_{y >= 0, sum(y) = z} ||y - x||^2
@mblondel
mblondel / seminb.py
Created Oct 28, 2015
Semi-supervised Naive Bayes
View seminb.py
# -*- coding: utf-8 -*-
# Copyright (C) 2010 Mathieu Blondel
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
@mblondel
mblondel / sparse_multiclass_numba.py
Last active Jan 27, 2020
Sparse Multiclass Classification in Numba!
View sparse_multiclass_numba.py
"""
(C) August 2013, Mathieu Blondel
# License: BSD 3 clause
This is a Numba-based reimplementation of the block coordinate descent solver
(without line search) described in the paper:
Block Coordinate Descent Algorithms for Large-scale Sparse Multiclass
Classification. Mathieu Blondel, Kazuhiro Seki, and Kuniaki Uehara.
Machine Learning, May 2013.
@mblondel
mblondel / projection_simplex_isotonic.py
Created Nov 2, 2019
Projection onto the probability simplex using isotonic regression.
View projection_simplex_isotonic.py
# Author: Mathieu Blondel
# License: BSD 3 clause
import numpy as np
from sklearn.isotonic import isotonic_regression
def projection_simplex(x, z=1):
"""
Compute argmin_{p : p >= 0 and \sum_i p_i = z} ||p - x||
@mblondel
mblondel / nmf_cd.py
Last active Jun 12, 2019
NMF by coordinate descent
View nmf_cd.py
"""
NMF by coordinate descent, designed for sparse data (without missing values)
"""
# Author: Mathieu Blondel <mathieu@mblondel.org>
# License: BSD 3 clause
import numpy as np
import scipy.sparse as sp
import numba
@mblondel
mblondel / matrix_sketch.py
Last active Feb 13, 2019
Frequent directions algorithm for matrix sketching.
View matrix_sketch.py
# (C) Mathieu Blondel, November 2013
# License: BSD 3 clause
import numpy as np
from scipy.linalg import svd
def frequent_directions(A, ell, verbose=False):
"""
Return the sketch of matrix A.
@mblondel
mblondel / fista.py
Created Nov 6, 2016
Efficient implementation of FISTA
View fista.py
"""
Efficient implementation of FISTA.
"""
# Author: Mathieu Blondel
# License: BSD 3 clause
import numpy as np