I have a similar open question here on Cross Validated (though not implementation focused, which I intend this question to be, so I think they are both valid).
I'm working on a project that uses sensors to monitor a persons GPS location. The coordinates will then be converted to a simple-grid representation. What I want to try and do is after recording a users routes, train a neural network to predict the next coordinates, i.e. take the example below where a user repeats only two routes over time, Home->A and Home->B.
I want to train an RNN/LSTM with sequences of varying lengths e.g. (14,3), (13,3), (12,3), (11,3), (10,3), (9,3), (8,3), (7,3), (6,3), (5,3), (4,3), (3,3), (2,3), (1,3) and then also predict with sequences of varying lengths e.g. for this example route if I called
route = [(14,3), (13,3), (12,3), (11,3), (10,3)] //pseudocode
pred = model.predict(route)
pred should give me (9,3) (or ideally even a longer prediction e.g. ((9,3), (8,3), (7,3), (6,3), (5,3), (4,3), (3,3), (2,3), (1,3))
How do I feed such training sequences to the init and forward operations identified below?
self.rnn = nn.RNN(input_size, hidden_dim, n_layers, batch_first=True)
out, hidden = self.rnn(x, hidden)
Also, should the entire route be a tensor or each set of coordinates within the route a tensor?
I'm not very experienced with RNNs, but I'll give it a try.
A few things to pay attention to before we start:
1. Your data is not normalized.
2. The output prediction you want (even after normalization) is not bounded to [-1, 1] range and therefore you cannot have tanh or ReLU activations acting on the output predictions.
To address your problem, I propose a recurrent net that given a current state (2D coordinate) predicts the next state (2D coordinates). Note that since this is a recurrent net, there is also a hidden state associated with each location. At first, the hidden state is zero, but as the net sees more steps, it updates its hidden state.
I propose a simple net to address your problem. It has a single RNN layer with 8 hidden states, and a fully connected layer on to to output the prediction.
class MyRnn(nn.Module):
def __init__(self, in_d=2, out_d=2, hidden_d=8, num_hidden=1):
super(MyRnn, self).__init__()
self.rnn = nn.RNN(input_size=in_d, hidden_size=hidden_d, num_layers=num_hidden)
self.fc = nn.Linear(hidden_d, out_d)
def forward(self, x, h0):
r, h = self.rnn(x, h0)
y = self.fc(r) # no activation on the output
return y, h
You can use your two sequences as training data, each sequence is a tensor of shape Tx1x2 where T is the sequence length, and each entry is two dimensional (x-y).
To predict (during training):
rnn = MyRnn()
pred, out_h = rnn(seq[:-1, ...], torch.zeros(1, 1, 8)) # given time t predict t+1
err = criterion(pred, seq[1:, ...]) # compare prediction to t+1
Once the model is trained, you can show it first k steps and continue to predict the next steps:
rnn.eval()
with torch.no_grad():
pred, h = rnn(s[:k,...], torch.zeros(1, 1, 8, dtype=torch.float))
# pred[-1, ...] is the predicted next step
prev = pred[-1:, ...]
for j in range(k+1, s.shape[0]):
pred, h = rnn(prev, h) # note how we keep track of the hidden state of the model. it is no longer init to zero.
prev = pred
I put everything together in a colab notebook so you can play with it.
For simplicity, I ignored the data normalization here, but you can find it in the colab notebook.
What's next?
These types of predictions are prone to error accumulation. This should be addressed during training, by shifting the inputs from the ground truth "clean" sequences to the actual predicted sequences, so the model will be able to compensate for its errors.
Related
I'm currently working on building an LSTM model to forecast time-series data using PyTorch. I used lag features to pass the previous n steps as inputs to train the network. I split the data into three sets, i.e., train-validation-test split, and used the first two to train the model. My validation function takes the data from the validation data set and calculates the predicted valued by passing it to the LSTM model using DataLoaders and TensorDataset classes. Initially, I've got pretty good results with R2 values in the region of 0.85-0.95.
However, I have an uneasy feeling about whether this validation function is also suitable for testing my model's performance. Because the function now takes the actual X values, i.e., time-lag features, from the DataLoader to predict y^ values, i.e., predicted target values, instead of using the predicted y^ values as features in the next prediction. This situation seems far from reality where the model has no clue of the real values of the previous time steps, especially if you forecast time-series data for longer time periods, say 3-6 months.
I'm currently a bit puzzled about tackling this issue and defining a function to predict future values relying on the model's values rather than the actual values in the test set. I have the following function predict, which makes a one-step prediction, but I haven't really figured out how to predict the whole test dataset using DataLoader.
def predict(self, x):
# convert row to data
x = x.to(device)
# make prediction
yhat = self.model(x)
# retrieve numpy array
yhat = yhat.to(device).detach().numpy()
return yhat
You can find how I split and load my datasets, my constructor for the LSTM model, and the validation function below. If you need more information, please do not hesitate to reach out to me.
Splitting and Loading Datasets
def create_tensor_datasets(X_train_arr, X_val_arr, X_test_arr, y_train_arr, y_val_arr, y_test_arr):
train_features = torch.Tensor(X_train_arr)
train_targets = torch.Tensor(y_train_arr)
val_features = torch.Tensor(X_val_arr)
val_targets = torch.Tensor(y_val_arr)
test_features = torch.Tensor(X_test_arr)
test_targets = torch.Tensor(y_test_arr)
train = TensorDataset(train_features, train_targets)
val = TensorDataset(val_features, val_targets)
test = TensorDataset(test_features, test_targets)
return train, val, test
def load_tensor_datasets(train, val, test, batch_size=64, shuffle=False, drop_last=True):
train_loader = DataLoader(train, batch_size=batch_size, shuffle=shuffle, drop_last=drop_last)
val_loader = DataLoader(val, batch_size=batch_size, shuffle=shuffle, drop_last=drop_last)
test_loader = DataLoader(test, batch_size=batch_size, shuffle=shuffle, drop_last=drop_last)
return train_loader, val_loader, test_loader
Class LSTM
class LSTMModel(nn.Module):
def __init__(self, input_dim, hidden_dim, layer_dim, output_dim, dropout_prob):
super(LSTMModel, self).__init__()
self.hidden_dim = hidden_dim
self.layer_dim = layer_dim
self.lstm = nn.LSTM(
input_dim, hidden_dim, layer_dim, batch_first=True, dropout=dropout_prob
)
self.fc = nn.Linear(hidden_dim, output_dim)
def forward(self, x, future=False):
h0 = torch.zeros(self.layer_dim, x.size(0), self.hidden_dim).requires_grad_()
c0 = torch.zeros(self.layer_dim, x.size(0), self.hidden_dim).requires_grad_()
out, (hn, cn) = self.lstm(x, (h0.detach(), c0.detach()))
out = out[:, -1, :]
out = self.fc(out)
return out
Validation (defined within a trainer class)
def validation(self, val_loader, batch_size, n_features):
with torch.no_grad():
predictions = []
values = []
for x_val, y_val in val_loader:
x_val = x_val.view([batch_size, -1, n_features]).to(device)
y_val = y_val.to(device)
self.model.eval()
yhat = self.model(x_val)
predictions.append(yhat.cpu().detach().numpy())
values.append(y_val.cpu().detach().numpy())
return predictions, values
I've finally found a way to forecast values based on predicted values from the earlier observations. As expected, the predictions were rather accurate in the short-term, slightly becoming worse in the long term. It is not so surprising that the future predictions digress over time, as they no longer depend on the actual values. Reflecting on my results and the discussions I had on the topic, here are my take-aways:
In real-life cases, the real values can be retrieved and fed into the model at each step of the prediction -be it weekly, daily, or hourly- so that the next step can be predicted with the actual values from the previous step. So, testing the performance based on the actual values from the test set may somewhat reflect the real performance of the model that is maintained regularly.
However, for predicting future values in the long term, forecasting, if you will, you need to make either multiple one-step predictions or multi-step predictions that span over the time period you wish to forecast.
Making multiple one-step predictions based on the values predicted the model yields plausible results in the short term. As the forecasting period increases, the predictions become less accurate and therefore less fit for the purpose of forecasting.
To make multiple one-step predictions and update the input after each prediction, we have to work our way through the dataset one by one, as if we are going through a for-loop over the test set. Not surprisingly, this makes us lose all the computational advantages that matrix operations and mini-batch training provide us.
An alternative could be predicting sequences of values, instead of predicting the next value only, say using RNNs with multi-dimensional output with many-to-many or seq-to-seq structure. They are likely to be more difficult to train and less flexible to make predictions for different time periods. An encoder-decoder structure may prove useful for solving this, though I have not implemented it by myself.
You can find the code for my function that forecasts the next n_steps based on the last row of the dataset X (time-lag features) and y (target value). To iterate over each row in my dataset, I would set batch_size to 1 and n_features to the number of lagged observations.
def forecast(self, X, y, batch_size=1, n_features=1, n_steps=100):
predictions = []
X = torch.roll(X, shifts=1, dims=2)
X[..., -1, 0] = y.item(0)
with torch.no_grad():
self.model.eval()
for _ in range(n_steps):
X = X.view([batch_size, -1, n_features]).to(device)
yhat = self.model(X)
yhat = yhat.to(device).detach().numpy()
X = torch.roll(X, shifts=1, dims=2)
X[..., -1, 0] = yhat.item(0)
predictions.append(yhat)
return predictions
The following line shifts values in the second dimension of the tensor by one so that a tensor [[[x1, x2, x3, ... , xn ]]] becomes [[[xn, x1, x2, ... , x(n-1)]]].
X = torch.roll(X, shifts=1, dims=2)
And, the line below selects the first element from the last dimension of the 3d tensor and sets that item to the predicted value stored in the NumPy ndarray (yhat), [[xn+1]]. Then, the new input tensor becomes [[[x(n+1), x1, x2, ... , x(n-1)]]]
X[..., -1, 0] = yhat.item(0)
Recently, I've decided to put together the things I had learned and the things I would have liked to know earlier. If you'd like to have a look, you can find the links down below. I hope you'll find it useful. Feel free to comment or reach out to me if you agree or disagree with any of the remarks I made above.
Building RNN, LSTM, and GRU for time series using PyTorch
Predicting future values with RNN, LSTM, and GRU using PyTorch
I would like to code with Keras a neural network that acts both as an autoencoder AND a classifier for semi-supervised learning. Take for example this dataset where there is a few labeled images and a lot of unlabeled images: https://cs.stanford.edu/~acoates/stl10/
Some papers listed here achieved that, or very similar things, successfully.
To sum up: if the model would have the same input data shape and the same "encoding" convolutional layers, but would split into two heads (fork-style), so there is a classification head and a decoding head, in a way that the unsupervised autoencoder will contribute to a good learning for the classification head.
With TensorFlow there would be no problem doing that as we have full control over the computational graph.
But with Keras, things are more high-level and I feel that all the calls to ".fit" must always provide all the data at once (so it would force me to tie together the classification head and the autoencoding head into one time-step).
One way in keras to almost do that would be with something that goes like this:
input = Input(shape=(32, 32, 3))
cnn_feature_map = sequential_cnn_trunk(input)
classification_predictions = Dense(10, activation='sigmoid')(cnn_feature_map)
autoencoded_predictions = decode_cnn_head_sequential(cnn_feature_map)
model = Model(inputs=[input], outputs=[classification_predictions, ])
model.compile(optimizer='rmsprop',
loss='binary_crossentropy',
metrics=['accuracy'])
model.fit([images], [labels, images], epochs=10)
However, I think and I fear that if I just want to fit things in that way it will fail and ask for the missing head:
for epoch in range(10):
# classifications step
model.fit([images], [labels, None], epochs=1)
# "semi-unsupervised" autoencoding step
model.fit([images], [None, images], epochs=1)
# note: ".train_on_batch" could probably be used rather than ".fit" to avoid doing a whole epoch each time.
How should one implement that behavior with Keras? And could the training be done jointly without having to split the two calls to the ".fit" function?
Sometimes when you don't have a label you can pass zero vector instead of one hot encoded vector. It should not change your result because zero vector doesn't have any error signal with categorical cross entropy loss.
My custom to_categorical function looks like this:
def tricky_to_categorical(y, translator_dict):
encoded = np.zeros((y.shape[0], len(translator_dict)))
for i in range(y.shape[0]):
if y[i] in translator_dict:
encoded[i][translator_dict[y[i]]] = 1
return encoded
When y contains labels, and translator_dict is a python dictionary witch contains labels and its unique keys like this:
{'unisex':2, 'female': 1, 'male': 0}
If an UNK label can't be found in this dictinary then its encoded label will be a zero vector
If you use this trick you also have to modify your accuracy function to see real accuracy numbers. you have to filter out all zero vectors from our metrics
def tricky_accuracy(y_true, y_pred):
mask = K.not_equal(K.sum(y_true, axis=-1), K.constant(0)) # zero vector mask
y_true = tf.boolean_mask(y_true, mask)
y_pred = tf.boolean_mask(y_pred, mask)
return K.cast(K.equal(K.argmax(y_true, axis=-1), K.argmax(y_pred, axis=-1)), K.floatx())
note: You have to use larger batches (e.g. 32) in order to prevent zero matrix update, because It can make your accuracy metrics crazy, I don't know why
Alternative solution
Use Pseudo Labeling :)
you can train jointly, you have to pass an array insted of single label.
I used fit_generator, e.g.
model.fit_generator(
batch_generator(),
steps_per_epoch=len(dataset) / batch_size,
epochs=epochs)
def batch_generator():
batch_x = np.empty((batch_size, img_height, img_width, 3))
gender_label_batch = np.empty((batch_size, len(gender_dict)))
category_label_batch = np.empty((batch_size, len(category_dict)))
while True:
i = 0
for idx in np.random.choice(len(dataset), batch_size):
image_id = dataset[idx][0]
batch_x[i] = load_and_convert_image(image_id)
gender_label_batch[i] = gender_labels[idx]
category_label_batch[i] = category_labels[idx]
i += 1
yield batch_x, [gender_label_batch, category_label_batch]
I am attempting to train an ANN on time series data in Keras. I have three vectors of data that are broken into scrolling window sequences (i.e. for vector l).
np.array([l[i:i+window_size] for i in range( len(l) - window_size)])
The target vector is similarly windowed so the neural net output is a prediction of the target vector for the next window_size number of time steps. All the data is normalized with a min-max scaler. It is fed into the neural network as a shape=(nb_samples, window_size, 3). Here is a plot of the 3 input vectors.
The only output I've managed to muster from the ANN is the following plot. Target vector in blue, predictions in red (plot is zoomed in to make the prediction pattern legible). Prediction vectors are plotted at window_size intervals so each one of the repeated patterns is one prediction from the net.
I've tried many different model architectures, number of epochs, activation functions, short and fat networks, skinny, tall. This is my current one (it's a little out there).
Conv1D(64,4, input_shape=(None,3)) ->
Conv1d(32,4) ->
Dropout(24) ->
LSTM(32) ->
Dense(window_size)
But nothing I try will affect the neural net from outputting this repeated pattern. I must be misunderstanding something about time-series or LSTMs in Keras. But I'm very lost at this point so any help is greatly appreciated. I've attached the full code at this repository.
https://github.com/jaybutera/dat-toy
I played with your code a little and I think I have a few suggestions for getting you on the right track. The code doesn't seem to match your graphs exactly, but I assume you've tweaked it a bit since then. Anyway, there are two main problems:
The biggest problem is in your data preparation step. You basically have the data shapes backwards, in that you have a single timestep of input for X and a timeseries for Y. Your input shape is (18830, 1, 8), when what you really want is (18830, 30, 8) so that the full 30 timesteps are fed into the LSTM. Otherwise the LSTM is only operating on one timestep and isn't really useful. To fix this, I changed the line in common.py from
X = X.reshape(X.shape[0], 1, X.shape[1])
to
X = windowfy(X, winsize)
Similarly, the output data should probably be only 1 value, from what I've gathered of your goals from the plotting function. There are certainly some situations where you want to predict a whole timeseries, but I don't know if that's what you want in this case. I changed Y_train to use fuels instead of fuels_w so that it only had to predict one step of the timeseries.
Training for 100 epochs might be way too much for this simple network architecture. In some cases when I ran it, it looked like there was some overfitting going on. Observing the decrease of loss in the network, it seems like maybe only 3-4 epochs are needed.
Here is the graph of predictions after 3 training epochs with the adjustments I mentioned. It's not a great prediction, but it looks like it's on the right track now at least. Good luck to you!
EDIT: Example predicting multiple output timesteps:
from sklearn import datasets, preprocessing
import numpy as np
from scipy import stats
from keras import models, layers
INPUT_WINDOW = 10
OUTPUT_WINDOW = 5 # Predict 5 steps of the output variable.
# Randomly generate some regression data (not true sequential data; samples are independent).
np.random.seed(11798)
X, y = datasets.make_regression(n_samples=1000, n_features=4, noise=.1)
# Rescale 0-1 and convert into windowed sequences.
X = preprocessing.MinMaxScaler().fit_transform(X)
y = preprocessing.MinMaxScaler().fit_transform(y.reshape(-1, 1))
X = np.array([X[i:i + INPUT_WINDOW] for i in range(len(X) - INPUT_WINDOW)])
y = np.array([y[i:i + OUTPUT_WINDOW] for i in range(INPUT_WINDOW - OUTPUT_WINDOW,
len(y) - OUTPUT_WINDOW)])
print(np.shape(X)) # (990, 10, 4) - Ten timesteps of four features
print(np.shape(y)) # (990, 5, 1) - Five timesteps of one features
# Construct a simple model predicting output sequences.
m = models.Sequential()
m.add(layers.LSTM(20, activation='relu', return_sequences=True, input_shape=(INPUT_WINDOW, 4)))
m.add(layers.LSTM(20, activation='relu'))
m.add(layers.RepeatVector(OUTPUT_WINDOW))
m.add(layers.LSTM(20, activation='relu', return_sequences=True))
m.add(layers.wrappers.TimeDistributed(layers.Dense(1, activation='sigmoid')))
print(m.summary())
m.compile(optimizer='adam', loss='mse')
m.fit(X[:800], y[:800], batch_size=10, epochs=60) # Train on first 800 sequences.
preds = m.predict(X[800:], batch_size=10) # Predict the remaining sequences.
print('Prediction:\n' + str(preds[0]))
print('Actual:\n' + str(y[800]))
# Correlation should be around r = .98, essentially perfect.
print('Correlation: ' + str(stats.pearsonr(y[800:].flatten(), preds.flatten())[0]))
I am messing around with LSTMs and have a conceptual question. I created a matrix of bogus data on the following rules:
For each 1-D list in the matrix:
If previous element is less than 10, then this next element is the previous one plus 1.
Else, this element is sin(previous element)
This way, it is a sequence that is pretty simply based on the previous information. I set up an LSTM to learn the recurrence and ran it to train on the lists one at a time. I have an LSTM layer followed by a fully connected feed-forward layer. It learns the +1 step very easily, but has trouble with the sin step. It will seemingly pick a random number between -1 and 1 when making the next element when the previous one was greater than 10. My question is this: is the training only modifying the variables in my fully connected feed forward layer? Is that why it can't learn the non-linear sin function?
Here's the code snippet in question:
lstm = rnn_cell.LSTMCell(lstmSize)
y_ = tf.placeholder(tf.float32, [None, OS])
outputs, state = rnn.rnn(lstm, x, dtype=tf.float32)
outputs = tf.transpose(outputs, [1, 0, 2])
last = tf.gather(outputs, int(outputs.get_shape()[0]) - 1)
weights = tf.Variable(tf.truncated_normal([lstmSize, OS]))
bias = tf.Variable(tf.constant(0.1, shape=[OS]))
y = tf.nn.elu(tf.matmul(last, weights) + bias)
error = tf.reduce_mean(tf.square(tf.sub(y_, y)))
train_step = tf.train.AdamOptimizer(learning_rate=1e-3).minimize(error)
The error and shape organization seems to be correct, at least in the sense that it does learn the +1 step quickly without crashing. Shouldn't the LSTM be able to handle the non-linear sin function? It seems almost trivially easy, so my guess is that I set something up wrong and the LSTM isn't learning anything.
I can't understand why dropout works like this in tensorflow. The blog of CS231n says that, "dropout is implemented by only keeping a neuron active with some probability p (a hyperparameter), or setting it to zero otherwise." Also you can see this from picture(Taken from the same site)
From tensorflow site, With probability keep_prob, outputs the input element scaled up by 1 / keep_prob, otherwise outputs 0.
Now, why the input element is scaled up by 1/keep_prob? Why not keep the input element as it is with probability and not scale it with 1/keep_prob?
This scaling enables the same network to be used for training (with keep_prob < 1.0) and evaluation (with keep_prob == 1.0). From the Dropout paper:
The idea is to use a single neural net at test time without dropout. The weights of this network are scaled-down versions of the trained weights. If a unit is retained with probability p during training, the outgoing weights of that unit are multiplied by p at test time as shown in Figure 2.
Rather than adding ops to scale down the weights by keep_prob at test time, the TensorFlow implementation adds an op to scale up the weights by 1. / keep_prob at training time. The effect on performance is negligible, and the code is simpler (because we use the same graph and treat keep_prob as a tf.placeholder() that is fed a different value depending on whether we are training or evaluating the network).
Let's say the network had n neurons and we applied dropout rate 1/2
Training phase, we would be left with n/2 neurons. So if you were expecting output x with all the neurons, now you will get on x/2. So for every batch, the network weights are trained according to this x/2
Testing/Inference/Validation phase, we dont apply any dropout so the output is x. So, in this case, the output would be with x and not x/2, which would give you the incorrect result. So what you can do is scale it to x/2 during testing.
Rather than the above scaling specific to Testing phase. What Tensorflow's dropout layer does is that whether it is with dropout or without (Training or testing), it scales the output so that the sum is constant.
Here is a quick experiment to disperse any remaining confusion.
Statistically the weights of a NN-layer follow a distribution that is usually close to normal (but not necessarily), but even in the case when trying to sample a perfect normal distribution in practice, there are always computational errors.
Then consider the following experiment:
DIM = 1_000_000 # set our dims for weights and input
x = np.ones((DIM,1)) # our input vector
#x = np.random.rand(DIM,1)*2-1.0 # or could also be a more realistic normalized input
probs = [1.0, 0.7, 0.5, 0.3] # define dropout probs
W = np.random.normal(size=(DIM,1)) # sample normally distributed weights
print("W-mean = ", W.mean()) # note the mean is not perfect --> sampling error!
# DO THE DRILL
h = defaultdict(list)
for i in range(1000):
for p in probs:
M = np.random.rand(DIM,1)
M = (M < p).astype(int)
Wp = W * M
a = np.dot(Wp.T, x)
h[str(p)].append(a)
for k,v in h.items():
print("For drop-out prob %r the average linear activation is %r (unscaled) and %r (scaled)" % (k, np.mean(v), np.mean(v)/float(k)))
Sample output:
x-mean = 1.0
W-mean = -0.001003985674840264
For drop-out prob '1.0' the average linear activation is -1003.985674840258 (unscaled) and -1003.985674840258 (scaled)
For drop-out prob '0.7' the average linear activation is -700.6128015029908 (unscaled) and -1000.8754307185584 (scaled)
For drop-out prob '0.5' the average linear activation is -512.1602655283492 (unscaled) and -1024.3205310566984 (scaled)
For drop-out prob '0.3' the average linear activation is -303.21194422742315 (unscaled) and -1010.7064807580772 (scaled)
Notice that the unscaled activations diminish due to the statistically imperfect normal distribution.
Can you spot an obvious correlation between the W-mean and the average linear activation means?
If you keep reading in cs231n, the difference between dropout and inverted dropout is explained.
In a network with no dropout, the activations in layer L will be aL. The weights of next layer (L+1) will be learned in such a manner that it receives aL and produces output accordingly. But with a network containing dropout (with keep_prob = p), the weights of L+1 will be learned in such a manner that it receives p*aL and produces output accordingly. Why p*aL? Because the Expected value, E(aL), will be probability_of_keeping(aL)*aL + probability_of_not_keeping(aL)*0 which will be equal to p*aL + (1-p)*0 = p*aL. In the same network, during testing time there will be no dropout. Hence the layer L+1 will receive aL simply. But its weights were trained to expect p*aL as input. Therefore, during testing time you will have to multiply the activations with p. But instead of doing this, you can multiply the activations with 1/p during training only. This is called inverted dropout.
Since we want to leave the forward pass at test time untouched (and tweak our network just during training), tf.nn.dropout directly implements inverted dropout, scaling the values.