I am working on a multi-class classification task using my own images.
filenames = [] # a list of filenames
labels = [] # a list of labels corresponding to the filenames
full_ds = tf.data.Dataset.from_tensor_slices((filenames, labels))
This full dataset will be shuffled and split into train, valid and test dataset
full_ds_size = len(filenames)
full_ds = full_ds.shuffle(buffer_size=full_ds_size*2, seed=128) # seed is used for reproducibility
train_ds_size = int(0.64 * full_ds_size)
valid_ds_size = int(0.16 * full_ds_size)
train_ds = full_ds.take(train_ds_size)
remaining = full_ds.skip(train_ds_size)
valid_ds = remaining.take(valid_ds_size)
test_ds = remaining.skip(valid_ds_size)
Now I am struggling to understand how each class is distributed in train_ds, valid_ds and test_ds. An ugly solution is to iterate all the element in the dataset and count the occurrence of each class. Is there any better way to solve it?
My ugly solution:
def get_class_distribution(dataset):
class_distribution = {}
for element in dataset.as_numpy_iterator():
label = element[1]
if label in class_distribution.keys():
class_distribution[label] += 1
else:
class_distribution[label] = 0
# sort dict by key
class_distribution = collections.OrderedDict(sorted(class_distribution.items()))
return class_distribution
train_ds_class_dist = get_class_distribution(train_ds)
valid_ds_class_dist = get_class_distribution(valid_ds)
test_ds_class_dist = get_class_distribution(test_ds)
print(train_ds_class_dist)
print(valid_ds_class_dist)
print(test_ds_class_dist)
The answer below assumes:
there are five classes.
labels are integers from 0 to 4.
It can be modified to suit your needs.
Define a counter function:
def count_class(counts, batch, num_classes=5):
labels = batch['label']
for i in range(num_classes):
cc = tf.cast(labels == i, tf.int32)
counts[i] += tf.reduce_sum(cc)
return counts
Use the reduce operation:
initial_state = dict((i, 0) for i in range(5))
counts = train_ds.reduce(initial_state=initial_state,
reduce_func=count_class)
print([(k, v.numpy()) for k, v in counts.items()])
A solution inspired by user650654 's answer, only using TensorFlow primitives (with tf.unique_with_counts instead of for loop):
In theory, this should have better performance and scale better to large datasets, batches or class count.
num_classes = 5
#tf.function
def count_class(counts, batch):
y, _, c = tf.unique_with_counts(batch[1])
return tf.tensor_scatter_nd_add(counts, tf.expand_dims(y, axis=1), c)
counts = train_ds.reduce(
initial_state=tf.zeros(num_classes, tf.int32),
reduce_func=count_class)
print(counts.numpy())
Similar and simpler version with numpy that actually had better performances for my simple use-case:
count = np.zeros(num_classes, dtype=np.int32)
for _, labels in train_ds:
y, _, c = tf.unique_with_counts(labels)
count[y.numpy()] += c.numpy()
print(count)
I am trying to train a triple loss model using a fit_generator. it requires three input and no output. so i have a function that generates hard triplets. the output from the triplets generator has a shape of (3,5,279) which is 3 inputs(anchor,positive and negative) for 5 batches and a total of 279 features. When i run the fit_generator it throws this error that "the list of Numpy arrays that you are passing to your model is not the size the model expected. Expected to see 3 array(s), but instead got the following list of 1 arrays" meanwhile i have passed a list of three arrays. the code is below. it works when i use the fit, however, i want to always call the generator function to generate my triplets as my batches. thanks in advance..this has taken me three days
def load_data():
path = "arrhythmia_data.txt"
f = open( path, "r")
data = []
#remove line breaker, comma separate and store in array
for line in f:
line = line.replace('\n','').replace('?','0')
line = line.split(",")
data.append(line)
f.close()
data = np.array(data).astype(np.float64)
#print(data.shape)
#create the class labels for input data
Y_train = data[:,-1:]
train = data[:,:-1]
normaliser = preprocessing.MinMaxScaler()
train = normaliser.fit_transform(train)
val = train[320:,:]
train = train[:320,:]
#create one hot encoding of the class labels of the data and separate them into train and test data
lb = LabelBinarizer()
encode = lb.fit_transform(Y_train)
nb_classes = int(len(encode[0]))
#one_hot_labels = keras.utils.to_categorical(labels, num_classes=10) this could also be used for one hot encoding
Y_val_e = encode[320:,:]
Y_train_e = encode[:320,:]
print(Y_train_e[0])
print(np.argmax(Y_train_e[0]))
val_in = []
train_in = []
#grouping and sorting the input data based on label id or name
for n in range(nb_classes):
images_class_n = np.asarray([row for idx,row in enumerate(train) if np.argmax(Y_train_e[idx])==n])
train_in.append(images_class_n)
images_class_n = np.asarray([row for idx,row in enumerate(val) if np.argmax(Y_val_e[idx])==n])
val_in.append(images_class_n)
#print(train_in[0].shape)
return train_in,val_in,Y_train_e,Y_val_e,nb_classes
train_in,val,Y_train,Y_val,nb_classes = load_data()
input_shape = (train_in[0].shape[1],)
def build_network(input_shape , embeddingsize):
'''
Define the neural network to learn image similarity
Input :
input_shape : shape of input images
embeddingsize : vectorsize used to encode our picture
'''
#in_ = Input(train.shape)
net = Sequential()
net.add(Dense(128, activation='relu', input_shape=input_shape))
net.add(Dense(128, activation='relu'))
net.add(Dense(256, activation='relu'))
net.add(Dense(4096, activation='sigmoid'))
net.add(Dense(embeddingsize, activation= None))
#Force the encoding to live on the d-dimentional hypershpere
net.add(Lambda(lambda x: K.l2_normalize(x,axis=-1)))
return net
class TripletLossLayer(Layer):
def __init__(self, alpha, **kwargs):
self.alpha = alpha
super(TripletLossLayer, self).__init__(**kwargs)
def triplet_loss(self, inputs):
anchor, positive, negative = inputs
p_dist = K.sum(K.square(anchor-positive), axis=-1)
n_dist = K.sum(K.square(anchor-negative), axis=-1)
return K.sum(K.maximum(p_dist - n_dist + self.alpha, 0), axis=0)
def call(self, inputs):
loss = self.triplet_loss(inputs)
self.add_loss(loss)
return loss
def build_model(input_shape, network, margin=0.2):
'''
Define the Keras Model for training
Input :
input_shape : shape of input images
network : Neural network to train outputing embeddings
margin : minimal distance between Anchor-Positive and Anchor-Negative for the lossfunction (alpha)
'''
# Define the tensors for the three input images
anchor_input = Input(input_shape, name="anchor_input")
positive_input = Input(input_shape, name="positive_input")
negative_input = Input(input_shape, name="negative_input")
# Generate the encodings (feature vectors) for the three images
encoded_a = network(anchor_input)
encoded_p = network(positive_input)
encoded_n = network(negative_input)
#TripletLoss Layer
loss_layer = TripletLossLayer(alpha=margin,name='triplet_loss_layer')([encoded_a,encoded_p,encoded_n])
# Connect the inputs with the outputs
network_train = Model(inputs=[anchor_input,positive_input,negative_input],outputs=loss_layer)
# return the model
return network_train
def get_batch_random(batch_size,s="train"):
# initialize result
triplets=[np.zeros((batch_size,m)) for i in range(3)]
for i in range(batch_size):
#Pick one random class for anchor
anchor_class = np.random.randint(0, nb_classes)
nb_sample_available_for_class_AP = X[anchor_class].shape[0]
#Pick two different random pics for this class => A and P. You can use same anchor as P if there is one one element for anchor
if nb_sample_available_for_class_AP<=1:
continue
[idx_A,idx_P] = np.random.choice(nb_sample_available_for_class_AP,size=2 ,replace=False)
#Pick another class for N, different from anchor_class
negative_class = (anchor_class + np.random.randint(1,nb_classes)) % nb_classes
nb_sample_available_for_class_N = X[negative_class].shape[0]
#Pick a random pic for this negative class => N
idx_N = np.random.randint(0, nb_sample_available_for_class_N)
triplets[0][i,:] = X[anchor_class][idx_A,:]
triplets[1][i,:] = X[anchor_class][idx_P,:]
triplets[2][i,:] = X[negative_class][idx_N,:]
return np.array(triplets)
def get_batch_hard(draw_batch_size,hard_batchs_size,norm_batchs_size,network,s="train"):
if s == 'train':
X = train_in
else:
X = val
#m, features = X[0].shape
#while True:
#Step 1 : pick a random batch to study
studybatch = get_batch_random(draw_batch_size,X)
#Step 2 : compute the loss with current network : d(A,P)-d(A,N). The alpha parameter here is omited here since we want only to order them
studybatchloss = np.zeros((draw_batch_size))
#Compute embeddings for anchors, positive and negatives
A = network.predict(studybatch[0])
P = network.predict(studybatch[1])
N = network.predict(studybatch[2])
#Compute d(A,P)-d(A,N)
studybatchloss = np.sum(np.square(A-P),axis=1) - np.sum(np.square(A-N),axis=1)
#Sort by distance (high distance first) and take the
selection = np.argsort(studybatchloss)[::-1][:hard_batchs_size]
#Draw other random samples from the batch
selection2 = np.random.choice(np.delete(np.arange(draw_batch_size),selection),norm_batchs_size,replace=False)
selection = np.append(selection,selection2)
triplets = [studybatch[0][selection,:], studybatch[1][selection,:],studybatch[2][selection,:]]
triplets = triplets.reshape(triplets.shape[0],triplets.shape[1],triplets.shape[2])
yield triplets
network = build_network(input_shape,embeddingsize=10)
hard = get_batch_hard(5,4,1,network,s="train")
network_train = build_model(input_shape,network)
optimizer = Adam(lr = 0.00006)
network_train.compile(loss=None,optimizer=optimizer)
#this works
#history = network_train.fit(hard,epochs=100,steps_per_epoch=1, verbose=2)
history = network_train.fit_generator(hard,epochs=10,steps_per_epoch=16, verbose=2)
# error:: the list of Numpy arrays that you are passing to your model is not the size the model
expected. Expected to see 3 array(s), but instead got the following list of 1 arrays:
I think that's beacause in your generator you are yielding the 3 inputs array in one list, you need to yield the 3 arrays independently:
triplet_1 = studybatch[0][selection,:]
triplet_2 = studybatch[1][selection,:]
triplet_3 = studybatch[2][selection,:]
yield [triplet_1, triplet_2, triplet_3]
I am currently trying to code the attention mechanism from this paper: "Effective Approaches to Attention-based Neural Machine Translation", Luong, Pham, Manning (2015). (I use global attention with the dot score).
However, I am unsure on how to input the hidden and output states from the lstm decode. The issue is that the input of the lstm decoder at time t depends on quantities that I need to compute using the output and hidden states from t-1.
Here is the relevant part of the code:
with tf.variable_scope('data'):
prob = tf.placeholder_with_default(1.0, shape=())
X_or = tf.placeholder(shape = [batch_size, timesteps_1, num_input], dtype = tf.float32, name = "input")
X = tf.unstack(X_or, timesteps_1, 1)
y = tf.placeholder(shape = [window_size,1], dtype = tf.float32, name = "label_annotation")
logits = tf.zeros((1,1), tf.float32)
with tf.variable_scope('lstm_cell_encoder'):
rnn_layers = [tf.nn.rnn_cell.LSTMCell(size) for size in [hidden_size, hidden_size]]
multi_rnn_cell = tf.nn.rnn_cell.MultiRNNCell(rnn_layers)
lstm_outputs, lstm_state = tf.contrib.rnn.static_rnn(cell=multi_rnn_cell,inputs=X,dtype=tf.float32)
concat_lstm_outputs = tf.stack(tf.squeeze(lstm_outputs))
last_encoder_state = lstm_state[-1]
with tf.variable_scope('lstm_cell_decoder'):
initial_input = tf.unstack(tf.zeros(shape=(1,1,hidden_size2)))
rnn_decoder_cell = tf.nn.rnn_cell.LSTMCell(hidden_size, state_is_tuple = True)
# Compute the hidden and output of h_1
for index in range(window_size):
output_decoder, state_decoder = tf.nn.static_rnn(rnn_decoder_cell, initial_input, initial_state=last_encoder_state, dtype=tf.float32)
# Compute the score for source output vector
scores = tf.matmul(concat_lstm_outputs, tf.reshape(output_decoder[-1],(hidden_size,1)))
attention_coef = tf.nn.softmax(scores)
context_vector = tf.reduce_sum(tf.multiply(concat_lstm_outputs, tf.reshape(attention_coef, (window_size, 1))),0)
context_vector = tf.reshape(context_vector, (1,hidden_size))
# compute the tilda hidden state \tilde{h}_t=tanh(W[c_t, h_t]+b_t)
concat_context = tf.concat([context_vector, output_decoder[-1]], axis = 1)
W_tilde = tf.Variable(tf.random_normal(shape = [hidden_size*2, hidden_size2], stddev = 0.1), name = "weights_tilde", trainable = True)
b_tilde = tf.Variable(tf.zeros([1, hidden_size2]), name="bias_tilde", trainable = True)
hidden_tilde = tf.nn.tanh(tf.matmul(concat_context, W_tilde)+b_tilde) # hidden_tilde is [1*64]
# update for next time step
initial_input = tf.unstack(tf.reshape(hidden_tilde, (1,1,hidden_size2)))
last_encoder_state = state_decoder
# predict the target
W_target = tf.Variable(tf.random_normal(shape = [hidden_size2, 1], stddev = 0.1), name = "weights_target", trainable = True)
logit = tf.matmul(hidden_tilde, W_target)
logits = tf.concat([logits, logit], axis = 0)
logits = logits[1:]
The part inside the loop is what I am unsure of. Does tensorflow remember the computational graph when I overwrite the variable "initial_input" and "last_encoder_state"?
I think your model will be much simplified if you use tf.contrib.seq2seq.AttentionWrapper with one of implementations: BahdanauAttention or LuongAttention.
This way it'll be possible to wire the attention vector on a cell level, so that cell output is already after attention applied. Example from the seq2seq tutorial:
cell = LSTMCell(512)
attention_mechanism = tf.contrib.seq2seq.LuongAttention(512, encoder_outputs)
attn_cell = tf.contrib.seq2seq.AttentionWrapper(cell, attention_mechanism, attention_size=256)
Note that this way you won't need a loop of window_size, because tf.nn.static_rnn or tf.nn.dynamic_rnn will instantiate the cells wrapped with attention.
Regarding your question: you should distinguish python variables and tensorflow graph nodes: you can assign last_encoder_state to a different tensor, the original graph node won't change because of this. This is flexible, but can be also misleading in the result network - you might think that you connect an LSTM to one tensor, but it's actually the other. In general, you shouldn't do that.
I'm trying to build a convolutional lstm autoencoder (that also predicts future and past) with Tensorflow, and it works to a certain degree, but the error sometimes jumps back up, so essentially, it never converges.
The model is as follows:
The encoder starts with a 64x64 frame from a 20 frame bouncing mnist video for each time step of the lstm. Every stacking layer of LSTM halfs it and increases the depth via 2x2 convolutions with a stride of 2. (so -->32x32x3 -->...--> 1x1x96)
On the other hand, the lstm performs 3x3 convolutions with a stride of 1 on its state. Both results are concatenated to form the new state. In the same way, the decoder uses transposed convolutions to go back to the original format. Then the squared error is calculated.
The error starts at around 2700 and it takes around 20 hours (geforce1060) to get down to ~1700. At which point the jumping back up (and it sometimes jumps back up to 2300 or even ridiculous values like 440300) happens often enough that I can't really get any lower. Also at that point, it can usually pinpoint where the number should be, but its too fuzzy to actually make out the digit...
I tried different learning rates and optimizers, so if anybody knows why that jumping happens, that'd make me happy :)
Here is a graph of the loss with epochs:
import tensorflow as tf
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
import os
os.environ["CUDA_DEVICE_ORDER"]="PCI_BUS_ID"
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
#based on code by loliverhennigh (Github)
class ConvCell(tf.contrib.rnn.RNNCell):
count = 0 #exists only to remove issues with variable scope
def __init__(self, shape, num_features, transpose = False):
self.shape = shape
self.num_features = num_features
self._state_is_tuple = True
self._transpose = transpose
ConvCell.count+=1
self.count = ConvCell.count
#property
def state_size(self):
return (tf.contrib.rnn.LSTMStateTuple(self.shape[0:4],self.shape[0:4]))
#property
def output_size(self):
return tf.TensorShape(self.shape[1:4])
#here comes to the actual conv lstm implementation, if transpose = true, it performs a deconvolution on the input
def __call__(self, inputs, state, scope=None):
with tf.variable_scope(scope or type(self).__name__+str(self.count)):
c, h = state
state_shape = h.shape
input_shape = inputs.shape
#filter variables and convolutions on data coming from the same cell, a time step previous
h_filters = tf.get_variable("h_filters",[3,3,state_shape[3],self.num_features])
h_filters_gates = tf.get_variable("h_filters_gates",[3,3,state_shape[3],3])
h_partial = tf.nn.conv2d(h,h_filters,[1,1,1,1],'SAME')
h_partial_gates = tf.nn.conv2d(h,h_filters_gates,[1,1,1,1],'SAME')
c_filters = tf.get_variable("c_filters",[3,3,state_shape[3],3])
c_partial = tf.nn.conv2d(c,c_filters,[1,1,1,1],'SAME')
#filters and convolutions/deconvolutions on data coming fromthe cell input
if self._transpose:
x_filters = tf.get_variable("x_filters",[2,2,self.num_features,input_shape[3]])
x_filters_gates = tf.get_variable("x_filters_gates",[2,2,3,input_shape[3]])
x_partial = tf.nn.conv2d_transpose(inputs,x_filters,[int(state_shape[0]),int(state_shape[1]),int(state_shape[2]),self.num_features],[1,2,2,1],'VALID')
x_partial_gates = tf.nn.conv2d_transpose(inputs,x_filters_gates,[int(state_shape[0]),int(state_shape[1]),int(state_shape[2]),3],[1,2,2,1],'VALID')
else:
x_filters = tf.get_variable("x_filters",[2,2,input_shape[3],self.num_features])
x_filters_gates = tf.get_variable("x_filters_gates",[2,2,input_shape[3],3])
x_partial = tf.nn.conv2d(inputs,x_filters,[1,2,2,1],'VALID')
x_partial_gates = tf.nn.conv2d(inputs,x_filters_gates,[1,2,2,1],'VALID')
#some more lstm gate business
gate_bias = tf.get_variable("gate_bias",[1,1,1,3])
h_bias = tf.get_variable("h_bias",[1,1,1,self.num_features*2])
gates = h_partial_gates + x_partial_gates + c_partial + gate_bias
i,f,o = tf.split(gates,3,axis=3)
#concatenate the units coming from the spacial and the temporal dimension to build a unified state
concat = tf.concat([h_partial,x_partial],3) + h_bias
new_c = tf.nn.relu(concat)*tf.sigmoid(i)+c*tf.sigmoid(f)
new_h = new_c * tf.sigmoid(o)
new_state = tf.contrib.rnn.LSTMStateTuple(new_c,new_h)
return new_h, new_state #its redundant, but thats how tensorflow likes it, apparently
#global variables
LEARNING_RATE = 0.005
ITERATIONS_PER_EPOCH = 80
BATCH_SIZE = 75
TEST = False #manual switch to go from training to testing
if TEST:
BATCH_SIZE = 1
inputs = tf.placeholder(tf.float32, (20, BATCH_SIZE, 64, 64,1))
shape0 = [BATCH_SIZE,64,64,2]
shape1 = [BATCH_SIZE,32,32,6]
shape2 = [BATCH_SIZE,16,16,12]
shape3 = [BATCH_SIZE,8,8,24]
shape4 = [BATCH_SIZE,4,4,48]
shape5 = [BATCH_SIZE,2,2,96]
shape6 = [BATCH_SIZE,1,1,192]
#apparently tf.multirnncell has very specific requirements for the initial states oO
initial_state1 = (tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape1),tf.zeros(shape1)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape2),tf.zeros(shape2)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape3),tf.zeros(shape3)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape4),tf.zeros(shape4)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape5),tf.zeros(shape5)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape6),tf.zeros(shape6)))
initial_state2 = (tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape5),tf.zeros(shape5)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape4),tf.zeros(shape4)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape3),tf.zeros(shape3)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape2),tf.zeros(shape2)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape1),tf.zeros(shape1)),tf.contrib.rnn.LSTMStateTuple(tf.zeros(shape0),tf.zeros(shape0)))
#encoding part of the autoencoder graph
cell1 = ConvCell(shape1,3)
cell2 = ConvCell(shape2,6)
cell3 = ConvCell(shape3,12)
cell4 = ConvCell(shape4,24)
cell5 = ConvCell(shape5,48)
cell6 = ConvCell(shape6,96)
mcell = tf.contrib.rnn.MultiRNNCell([cell1,cell2,cell3,cell4,cell5,cell6])
rnn_outputs, rnn_states = tf.nn.dynamic_rnn(mcell, inputs[0:20,:,:,:],initial_state=initial_state1,dtype=tf.float32, time_major=True)
#decoding part of the autoencoder graph, forward block and backwards block
cell9a = ConvCell(shape5,48,transpose = True)
cell10a = ConvCell(shape4,24,transpose = True)
cell11a = ConvCell(shape3,12,transpose = True)
cell12a = ConvCell(shape2,6,transpose = True)
cell13a = ConvCell(shape1,3,transpose = True)
cell14a = ConvCell(shape0,1,transpose = True)
mcella = tf.contrib.rnn.MultiRNNCell([cell9a,cell10a,cell11a,cell12a,cell13a,cell14a])
cell9b = ConvCell(shape5,48,transpose = True)
cell10b = ConvCell(shape4,24,transpose = True)
cell11b= ConvCell(shape3,12,transpose = True)
cell12b = ConvCell(shape2,6,transpose = True)
cell13b = ConvCell(shape1,3,transpose = True)
cell14b = ConvCell(shape0,1,transpose = True)
mcellb = tf.contrib.rnn.MultiRNNCell([cell9b,cell10b,cell11b,cell12b,cell13b,cell14b])
def PredictionLayer(rnn_outputs,viewPoint = 11, reverse = False):
predLength = viewPoint-2 if reverse else 20-viewPoint #vision is the input for the decoder
vision = tf.concat([rnn_outputs[viewPoint-1:viewPoint,:,:,:],tf.zeros([predLength,BATCH_SIZE,1,1,192])],0)
if reverse:
rnn_outputs2, rnn_states = tf.nn.dynamic_rnn(mcellb, vision, initial_state = initial_state2, time_major=True)
else:
rnn_outputs2, rnn_states = tf.nn.dynamic_rnn(mcella, vision, initial_state = initial_state2, time_major=True)
mean = tf.reduce_mean(rnn_outputs2,4)
if TEST:
return mean
if reverse:
return tf.reduce_sum(tf.square(mean-inputs[viewPoint-2::-1,:,:,:,0]))
else:
return tf.reduce_sum(tf.square(mean-inputs[viewPoint-1:20,:,:,:,0]))
if TEST:
mean = tf.concat([PredictionLayer(rnn_outputs,11,True)[::-1,:,:,:],createPredictionLayer(rnn_outputs,11)],0)
else: #training part of the graph
error = tf.zeros([1])
for i in range(8,15): #range size of 7 or less works, 9 or more does not, no idea why
error += PredictionLayer(rnn_outputs, i)
error += PredictionLayer(rnn_outputs, i, True)
train_fn = tf.train.RMSPropOptimizer(learning_rate=LEARNING_RATE).minimize(error)
################################################################################
## TRAINING LOOP ##
################################################################################
#code based on siemanko/tf_lstm.py (Github)
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.8)
saver = tf.train.Saver(restore_sequentially=True, allow_empty=True,)
session = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
session.run(tf.global_variables_initializer())
vids = np.load("mnist_test_seq.npy") #20/10000/64/64 , moving mnist dataset from http://www.cs.toronto.edu/~nitish/unsupervised_video/
vids = vids[:,0:6000,:,:] #training set
saver.restore(session,tf.train.latest_checkpoint('./conv_lstm_multiples_v2/'))
#saver.restore(session,'.\conv_lstm_multiples\iteration-74')
for epoch in range(1000):
if TEST:
break
epoch_error = 0
#randomize batches each epoch
vids = np.swapaxes(vids,0,1)
np.random.shuffle(vids)
vids = np.swapaxes(vids,0,1)
for i in range(ITERATIONS_PER_EPOCH):
#running the graph and feeding data
err,_ = session.run([error, train_fn], {inputs: np.expand_dims(vids[:,i*BATCH_SIZE:(i+1)*BATCH_SIZE,:,:],axis=4)})
print(err)
epoch_error += err
#training error each epoch and regular saving
epoch_error /= (ITERATIONS_PER_EPOCH*BATCH_SIZE*4096*20*7)
if (epoch+1) % 5 == 0:
saver.save(session,'.\conv_lstm_multiples_v2\iteration',global_step=epoch)
print("saved")
print("Epoch %d, train error: %f" % (epoch, epoch_error))
#testing
plt.ion()
f, axarr = plt.subplots(2)
vids = np.load("mnist_test_seq.npy")
for i in range(6000,10000):
img = session.run([mean], {inputs: np.expand_dims(vids[:,i:i+1,:,:],axis=4)})
for j in range(20):
axarr[0].imshow(img[0][j,0,:,:])
axarr[1].imshow(vids[j,i,:,:])
plt.show()
plt.pause(0.1)
Usually this happens when gradients' magnitude is very high at some point and causes your network parameters to change a lot. To verify that it is indeed the case, you can produce the same plot of gradient magnitudes and see if they jump right before the loss jump. Assuming this is the case, the classic approach is to use gradient clipping (or go all the way to natural gradient).
I want to design a single layer RNN in Tensorflow such that last output (y(t-1)) is participated in updating the hidden state.
h(t) = tanh(W_{ih} * x(t) + W_{hh} * h(t) + **W_{oh}y(t - 1)**)
y(t) = W_{ho}*h(t)
How can I feed last input y(t - 1) as input for updating the hidden state?
Is y(t-1) the last input or output? In both cases it is not a straight fit with the TensorFlow RNN cell abstraction. If your RNN is simple you can just write the loop on your own, then you have full control. Another way that I would use is to pre-process your RNN input, e.g., do something like:
processed_input[t] = tf.concat(input[t], input[t-1])
Then call the RNN cell with processed_input and split there.
One possibility is to use tf.nn.raw_rnn which I found in this article. Check my answer to this related post.
I would call what you described an "autoregressive RNN". Here's an (incomplete) code snippet that shows how you can create one using tf.nn.raw_rnn:
import tensorflow as tf
LSTM_SIZE = 128
BATCH_SIZE = 64
HORIZON = 10
lstm_cell = tf.nn.rnn_cell.LSTMCell(LSTM_SIZE, use_peepholes=True)
class RnnLoop:
def __init__(self, initial_state, cell):
self.initial_state = initial_state
self.cell = cell
def __call__(self, time, cell_output, cell_state, loop_state):
emit_output = cell_output # == None for time == 0
if cell_output is None: # time == 0
initial_input = tf.fill([BATCH_SIZE, LSTM_SIZE], 0.0)
next_input = initial_input
next_cell_state = self.initial_state
else:
next_input = cell_output
next_cell_state = cell_state
elements_finished = (time >= HORIZON)
next_loop_state = None
return elements_finished, next_input, next_cell_state, emit_output, next_loop_state
rnn_loop = RnnLoop(initial_state=initial_state_tensor, cell=lstm_cell)
rnn_outputs_tensor_array, _, _ = tf.nn.raw_rnn(lstm_cell, rnn_loop)
rnn_outputs_tensor = rnn_outputs_tensor_array.stack()
Here we initialize internal state of LSTM with some vector initial_state_tensor, and feed zero array as input at t=0. After that, the output of the current timestep is the input for the next timestep.