def create_toy_data(N=1000, n_trials=1):
# Create Gabor kernel
rfsize = (15, 15)
K_true = createRF(name='gabor',
size=rfsize,
threshold=0.2,
dtype=np.float64,
frequency=0.5,
sigma=2 * [0.35])
K_true[K_true < 0] *= 1.25
# GWN Stimulus
X = np.random.randn(N, rfsize[0] * rfsize[1])
# Simulate response of LNP model
cell = SimpleCell(K_true,
threshold=1.5,
stddev=0.5,
n_trials=n_trials,
rectify=True)
Y = cell.simulate(X).astype(np.float64)
print "%d spikes (p(spike) = %0.3f)" % (np.sum(Y > 0), np.mean(Y > 0))
return X, Y, K_true
def create_toy_data(N=1000, n_trials=1):
# Create Gabor kernel
rfsize = (15, 15)
K_true = createRF(name='gabor', size=rfsize, threshold=0.2,
dtype=np.float64, frequency=0.5, sigma=2*[0.35])
K_true[K_true < 0] *= 1.25
# GWN Stimulus
X = np.random.randn(N, rfsize[0] * rfsize[1])
# Simulate response of LNP model
cell = SimpleCell(K_true, threshold=1.5, stddev=0.5, n_trials=n_trials,
rectify=True)
Y = cell.simulate(X).astype(np.float64)
print "%d spikes (p(spike) = %0.3f)" % (np.sum(Y > 0),
np.mean(Y > 0))
return X, Y, K_true
def create_toy_data(N=1000, nonlin_order=3):
# Create Gabor kernel
rfsize = (25, 25)
K_true = createRF(name='gabor', size=rfsize, threshold=0.2,
dtype=np.float64, frequency=0.5, sigma=2*[0.35])
# GWN Stimulus)
X = createGratings(size=rfsize, N=N, center=True, whiten=True)
# Poisson spike generation
z = np.dot(X, K_true.ravel())
z[z < 0] = 0
z = z ** nonlin_order
z /= z.max()
Y = (z > np.random.rand(N)).astype(np.float64)
print "%d spikes (p(spike) = %0.3f)" % (np.sum(Y > 0),
np.mean(Y > 0))
return X, Y, K_true
def create_toy_data(N=1000, nonlin_order=3):
# Create Gabor kernel
rfsize = (25, 25)
K_true = createRF(name='gabor',
size=rfsize,
threshold=0.2,
dtype=np.float64,
frequency=0.5,
sigma=2 * [0.35])
# GWN Stimulus)
X = createGratings(size=rfsize, N=N, center=True, whiten=True)
# Poisson spike generation
z = np.dot(X, K_true.ravel())
z[z < 0] = 0
z = z**nonlin_order
z /= z.max()
Y = (z > np.random.rand(N)).astype(np.float64)
print "%d spikes (p(spike) = %0.3f)" % (np.sum(Y > 0), np.mean(Y > 0))
return X, Y, K_true
