Wieland Brendel wielandbrendel

wielandbrendel / brendel_bethge.py

Created October 27, 2019 19:40

BrendelBethge attacks

	import logging
	from foolbox.utils import onehot_like

	from foolbox.attacks.base import Attack
	from foolbox.attacks.base import generator_call_decorator

	import numpy as np
	from numba import jitclass

	import line_profiler

wielandbrendel / numba_closure.py

Last active September 24, 2019 13:25

wielandbrendel / AE-Square-Stimuli.ipynb

Created June 29, 2016 16:35

toy-vae

Sorry, something went wrong. Reload?

Sorry, we cannot display this file.

Sorry, this file is invalid so it cannot be displayed.

wielandbrendel / vgg19.py

Created March 7, 2016 14:47

Loading of VGG-19

	def get_vgg19(file='vgg19.pkl', leak=0, only_mean=False):

	if leak == 0:
	rectify = lasagne.nonlinearities.rectify
	else:
	rectify = LeakyRectify(leak)

	def build_model():
	net = {}
	net['input'] = InputLayer((None, 3, 224, 224))

wielandbrendel / rec_network_sim.py

Created September 30, 2015 12:09

Recurrent network simulation in Cython

	cimport cython
	cimport numpy as np
	from libc.math cimport log, exp, sqrt, sin

	import numpy as np

	@cython.boundscheck(False)
	@cython.wraparound(False)
	@cython.cdivision(True)
	def c_analytic_simulate(np.ndarray[double, ndim=2] A, np.ndarray[double, ndim=2] T, np.ndarray[double, ndim=2] Of,

wielandbrendel / recurrent_simulation.jl

Last active September 30, 2015 11:33

Updated simulation code (no devec + inplace BLAS operations)

	blas_set_num_threads(1)

	function train_network(A, T, Of, cs, dt)
	I, N = size(T)
	z = zeros(I)
	r = zeros(N)
	Az = Array(Float64, I)
	Ofr = Array(Float64, N)
	I_teach = Array(Float64, N)
	Tz = Array(Float64, N)

wielandbrendel / gist:ccf1ff6f8f92139439be

Last active September 3, 2015 13:12

Small CNN trained on CIFAR 10 using Keras

	import os
	os.environ['THEANO_FLAGS'] = 'device=gpu0, assert_no_cpu_op=raise, on_unused_input=ignore, floatX=float32'

	from __future__ import absolute_import
	from __future__ import print_function
	from keras.datasets import cifar10
	from keras.preprocessing.image import ImageDataGenerator
	from keras.models import Sequential
	from keras.layers.core import Dense, Dropout, Activation, Flatten
	from keras.layers.convolutional import Convolution2D, MaxPooling2D

wielandbrendel / gist:8505669e9a54b0446ce5

Created February 16, 2015 16:14

Using kernel_stretch from random.hpp from cudarray leaves input array unchanged

wielandbrendel / gist:e999c85f834dc2bd2cec

Created February 16, 2015 16:09

Simple Cuda kernel to multiply array with a constant (plus offset)

	#include <stdio.h>

	#define N 10
	#define kNumBlockThreads 512

	// copied from random.hpp
	#define CUDA_GRID_STRIDE_LOOP(i, n) \
	for (int i = blockIdx.x * blockDim.x + threadIdx.x; \
	i < n; \
	i += blockDim.x * gridDim.x)