Need help with Variational-Lstm-Autoencoder?
Click the “chat” button below for chat support from the developer who created it, or find similar developers for support.

About the developer

174 Stars 40 Forks MIT License 27 Commits 9 Opened issues


Lstm variational auto-encoder for time series anomaly detection and features extraction

Services available


Need anything else?

Contributors list

No Data


CI status

Variational auto-encoder for anomaly detection/features extraction, with lstm cells (stateless or stateful).



$ pip install --upgrade git+


from LstmVAE import LSTM_Var_Autoencoder
from LstmVAE import preprocess

preprocess(df) #return normalized df, check NaN values replacing it with 0

df = df.reshape(-1,timesteps,n_dim) #use 3D input, n_dim = 1 for 1D time series.

vae = LSTM_Var_Autoencoder(intermediate_dim = 15,z_dim = 3, n_dim=1, stateful = True) #default stateful = False, learning_rate=0.001, batch_size = 100, num_epochs = 200, opt = tf.train.AdamOptimizer, REG_LAMBDA = 0.01, grad_clip_norm=10, optimizer_params=None, verbose = True)

"""REG_LAMBDA is the L2 loss lambda coefficient, should be set to 0 if not desired. optimizer_param : pass a dict = {} """

x_reconstructed, recons_error = vae.reconstruct(df, get_error = True) #returns squared error

x_reduced = vae.reduce(df) #latent space representation


Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.




Tutorial on variational Autoencoders

A Multimodal Anomaly Detector for Robot-Assisted Feeding Using an LSTM-based Variational Autoencoder

Variational Autoencoder based Anomaly Detection using Reconstruction Probability

The Generalized Reparameterization Gradient

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.