def print_posterior(Z, W, data_dim):
N, K = Z.shape
for j in range(2**K):
binary = list(map(int, "{0:b}".format(j)))
pad_binary = [0] * (K - len(binary)) + binary
prob = np.exp(Z_posterior(pad_binary, Z))
if prob > 0.01:
pad_binary = np.array(pad_binary)
reconstruct = np.dot(pad_binary, W)
print("pad binary, reconstruct, probability")
print(pad_binary)
print(prob)
display_W(reconstruct, data_dim, 'four')
def print_paintbox(tree, W, data_dim, flag='four'):
small_x, small_y, big_x, big_y = data_dim
vec = get_vec(tree)
F, D = get_FD(tree)
K = int(math.log(D, 2))
print("outputting paintbox")
for j in range(D):
binary = list(map(int, "{0:b}".format(j)))
pad_binary = [0] * (K - len(binary)) + binary
prob = Z_paintbox(pad_binary, vec)
if prob > 0.01:
pad_binary = np.array(pad_binary)
reconstruct = np.dot(pad_binary, W)
print("pad binary, reconstruct, probability")
print(pad_binary)
print(prob)
display_W(reconstruct, data_dim, flag)
return 0
data_type,
corr_value=2)
Y = full_data[:data_count, :]
held_out = full_data[data_count:, :]
Z, W, ll_set, pred_ll = ugibbs_sampler(Y, held_out, alpha, sig, sig_w,
iterate, select)
approx = np.dot(Z, W)
for i in range(10):
print("sample: " + str(i))
print("features probability")
#print(prob_matrix[i,:])
print("features selected")
print(Z[i, :])
display_W(approx[i:i + 1, :], data_dim, 'nine')
print("data: " + str(i))
display_W(Y[i:i + 1, :], data_dim, 'nine')
#Output the posterior
Z_trunc, W_trunc = truncate(Z, W, select)
print_posterior(Z_trunc, W_trunc)
# display_W(W)
# Z_final = Z_set[iterate-1]
# plotRowHistogram(Z_final)
# plotMatrixHistogram(Z_final)
# title = 'Slice Sampler: log likelihood vs. iterations'
# x_axis = 'iterations'
# y_axis = 'log likelihood'
# data_x = [i for i in range(1,iterate+1)]
# data_y = ll_set
title = 'Nonparametric Paintbox: feature vs. time'
x_axis = 'time'
y_axis = 'features'
data_x = time_avg
data_y = f_avg
plot(title, x_axis, y_axis, data_x, data_y)
#visual verification of data reconstruction
approx = np.dot(Z, W)
for i in range(10):
print("sample: " + str(i))
print("features probability")
print(prob_matrix[i, :])
print("features selected")
print(Z[i, :])
display_W(approx[i:i + 1, :], small_x, small_y, big_x, big_y, flag)
print("data: " + str(i))
display_W(Y[i:i + 1, :], small_x, small_y, big_x, big_y, flag)
#printing paintbox (only makes sense for single run)
print_paintbox(tree, W, small_x, small_y, big_x, big_y, flag)
#display_W(W,small_x,small_y,big_x,big_y,'nine')
# pb_scale = scale(pb)
# plt.imshow(pb_scale,interpolation='nearest')
# plt.show()
#print(ll_list)
#plt.imshow(W,interpolation='nearest')
#plt.show()
#display_W(W)
D = res**F #discretization
T = 36 #length of datapoint
N = 100 #data size
sig = 0.1
sig_w = 5.0
print("GENERATE DATA")
Y,Z_gen,gen_pb = generate_data(res,D,F,N,T,sig)
print('FINISH GENERATE')
#gen_pb = scale(gen_pb)
#plt.imshow(gen_pb,interpolation='nearest')
#plt.show()
iterate = 100
#print("DATA")
#print(Y)
#init variables
#profile.run('gibbs_sample(Y,sig,iterate,D,F,N,T)')
ll_list,Z,W,pb = gibbs_sample(Y,sig,sig_w,iterate,D,F,N,T)
approx = np.dot(Z,W)
#print(Z)
pb_scale = scale(pb)
plt.imshow(pb_scale,interpolation='nearest')
plt.show()
#print(ll_list)
#plt.imshow(W,interpolation='nearest')
#plt.show()
display_W(W)
plt.plot(ll_list)
plt.show()
data_type = 'random'
#full_data,Z_gen = generate_data(feature_count,data_count + held_out,T,sig,data_type)
full_data, Z_gen = construct_data(small_x,
small_y,
big_x,
big_y,
data_count + held_out,
sig,
data_type,
corr_value=2)
Y = full_data[:data_count, :]
held_out = full_data[data_count:, :]
sig_alg = 0.1
nu_set, phi_set, Phi_set, tau_set, pred_ll = run_vi(
Y, held_out, alpha, sig_w, sig_alg, iterate, feature_count)
W = phi_set[iterate - 1]
display_W(W, small_x, small_y, big_x, big_y, 'nine')
print(pred_ll)
#A = phi_set[34]
# print(elbo_set)
# nu = nu_set[iterate-1]
# print(nu)
# print(Phi_set[iterate-1])
# N,K = nu.shape
# nu_var = np.zeros((N,K))
# for n in range(N):
# for k in range(K):
# nu_var[n,k] = nu[n,k]*(1-nu[n,k])
# variance = nu_var.sum()
# print(variance)
def upaintbox_sample(data_dim, log_res, hold, Y, held_out, ext, sig, sig_w,
iterate, K, truncate):
small_x, small_y, big_x, big_y = data_dim
N, T = Y.shape
#technically this is a bug
#generating tree with res = 1 is wrong.
#but you fixed it in tree paintbox hard coded 0.5
res = 1
tree = gen_tree(K, res)
ctree, ptree = tree
Z = draw_Z_tree(tree, N)
#Z = np.loadtxt('assignments.txt')
print(Z)
#W = sample_W(Y,Z,sig,sig_w)
W = np.reshape(np.random.normal(0, sig_w, K * T), (K, T))
#W = np.loadtxt('features.txt')
# full = generate_gg_blocks()
# W = np.zeros((3,T))
# W[0,:] = full[0,:]
# W[1,:] = full[2,:]
# W[2,:] = full[0,:] + full[2,:]
display_W(W, 'four')
ll_list = []
iter_time = []
f_count = []
lapse_data = []
pred_ll = []
pred = 0
rec = 0
for redo in range(1):
if redo == 1:
res = 1
N, K = Z.shape
tree = gen_tree(K, res)
ctree, ptree = tree
for it in range(iterate):
if it % hold == 0:
if res < 2**log_res:
res = res * 2
start = time.time()
N, K = Z.shape
#sample Z
Z, prob_matrix = sample_Z(Y, Z, W, sig, sig_w, tree)
if it % 10 == 0:
print("iteration: " + str(it))
print("Sparsity: " + str(np.sum(Z, axis=0)))
print('predictive log likelihood: ' + str(pred))
print('recover log likelihood: ' + str(rec))
#sample paintbox
tree, lapse = sample_pb(Z, tree, res)
#sample W
W = sample_W(Y, Z, sig, sig_w)
#add new features
ll_list.append(log_data_zw(Y, Z, W, sig))
F, D = get_FD(tree)
f_count.append(F)
#predictive log likelihood
if it % 100 == 0:
pred = pred_ll_paintbox(held_out, W, tree, sig)
pred_ll.append(pred)
rec = recover_paintbox(held_out, held_out[:, :T / 2], W, tree,
sig)
if it % 500 == 0 and it > 0:
print_paintbox(tree, W, data_dim, 'four')
#if it%200 == 0 and it > 0:
# display_W(W,data_dim,'nine')
#handling last iteration edge case
drop = 0
if it == iterate - 1:
drop = 1
Z, W, tree = new_feature(Y, Z, W, tree, ext, K, res, sig, sig_w,
drop, truncate)
end = time.time()
iter_time.append(end - start)
lapse_data.append(lapse)
#iter_time = np.cumsum(iter_time)
return (ll_list, iter_time, f_count, lapse_data, Z, W, prob_matrix,
pred_ll, tree)
title = 'Nonparametric Paintbox: feature vs. time'
x_axis = 'time'
y_axis = 'features'
data_x = time_avg
data_y = f_avg
plot(title, x_axis, y_axis, data_x, data_y)
#visual verification of data reconstruction
approx = np.dot(Z, W)
for i in range(10):
print("sample: " + str(i))
print("features probability")
print(prob_matrix[i, :])
print("features selected")
print(Z[i, :])
display_W(approx[i:i + 1, :], data_dim, flag)
print("data: " + str(i))
display_W(Y[i:i + 1, :], data_dim, flag)
#printing paintbox (only makes sense for single run)
print_paintbox(tree, W, data_dim, flag)
#display_W(W,small_x,small_y,big_x,big_y,'nine')
# pb_scale = scale(pb)
# plt.imshow(pb_scale,interpolation='nearest')
# plt.show()
#print(ll_list)
#plt.imshow(W,interpolation='nearest')
#plt.show()
#display_W(W)
