parser.add_argument('--exp', nargs='+', default=[], help="Nombre de los experimentos") args = parser.parse_args() lexperiments = args.exp if not args.batch: # 'e150514''e120503''e110616''e150707''e151126''e120511' lexperiments = ['e150514'] for exp in lexperiments: datainfo = experiments[exp] for sensor in datainfo.sensors: print(sensor) for dfile in [datainfo.datafiles[0]]: print(dfile) f = datainfo.open_experiment_data(mode='r') data = datainfo.get_peaks_resample_PCA(f, dfile, sensor) for d in data: mn = np.mean(d[0:20]) print d[0:20] plotListSignals([d]) datainfo.close_experiment_data(f)
datamat = datainfo.get_raw_data(f, dfile) for i in range(len(vecsync)): saverage = np.zeros((len(lsensors), (wlen*2))) scounts = np.zeros(len(lsensors)) for syn in vecsync[i]: stime = time_sync_sensor(syn, nsensor) for j in range(len(lsensors)): saverage[j] += datamat[stime-wlen:stime+wlen,j] # for s in syn: # saverage[s[0]] += datamat[stime-wlen:stime+wlen,s[0]] # scounts[s[0]] += 1 for i in range(len(lsensors)): saverage[i] /= len(vecsync[i]) # for i in range(len(lsensors)): # saverage[i] /= scounts[i] plotListSignals(saverage, ncols=2) datainfo.close_experiment_data(f) # for s in lsynchs_pruned: # print s
# construct the stacked denoising autoencoder class sda = SdA( numpy_rng=numpy_rng, n_ins=data.shape[1], hidden_layers_sizes=[1000, 500, 20, data.shape[1]], n_outs=10 ) # end-snippet-3 start-snippet-4 pretraining_fns = sda.pretraining_functions(train_set_x=train_set_x, batch_size=batch_size) print('... pre-training the model') start_time = timeit.default_timer() ## Pre-train layer-wise corruption_levels = [.1, .1, .1, .1] for i in range(sda.n_layers): # go through pretraining epochs for epoch in range(pretraining_epochs): # go through the training set c = [] for batch_index in range(n_train_batches): c.append(pretraining_fns[i](index=batch_index, corruption=corruption_levels[i], lr=pretrain_lr)) print('Pre-training layer %i, epoch %d, cost %f' % (i, epoch, numpy.mean(c))) end_time = timeit.default_timer() print(sda.dA_layers[-1].W.get_value(borrow=True).shape) plotListSignals(sda.dA_layers[-1].W.get_value(borrow=True), ncols=5)
cost, updates = da.get_cost_updates( corruption_level=0.1, learning_rate=learning_rate ) train_da = theano.function( [index], cost, updates=updates, givens={ x: train_set_x[index * batch_size: (index + 1) * batch_size] } ) start_time = timeit.default_timer() # go through training epochs for epoch in range(training_epochs): # go through trainng set c = [] for batch_index in range(n_train_batches): c.append(train_da(batch_index)) print('Training epoch %d, cost ' % epoch, numpy.mean(c)) end_time = timeit.default_timer() print(da.W.get_value(borrow=True).shape) plotListSignals(da.W.get_value(borrow=True).T, ncols=5)
da = dA(numpy_rng=rng, theano_rng=theano_rng, input=x, n_visible=data.shape[1], n_hidden=20) cost, updates = da.get_cost_updates(corruption_level=0.1, learning_rate=learning_rate) train_da = theano.function( [index], cost, updates=updates, givens={x: train_set_x[index * batch_size:(index + 1) * batch_size]}) start_time = timeit.default_timer() # go through training epochs for epoch in range(training_epochs): # go through trainng set c = [] for batch_index in range(n_train_batches): c.append(train_da(batch_index)) print('Training epoch %d, cost ' % epoch, numpy.mean(c)) end_time = timeit.default_timer() print(da.W.get_value(borrow=True).shape) plotListSignals(da.W.get_value(borrow=True).T, ncols=5)