Implicit Rank-Minimizing Autoencoder

Authors: Li Jing, Jure Zbontar, yann lecun

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically demonstrate IRMAE s regularization behavior through a synthetic dataset and show that it learns good representation with a much smaller latent dimension. Then we demonstrate superior representation learning performance of our method against a standard deterministic autoencoder and comparable performance to a variational autoencoder on MNIST dataset and Celeb A dataset through a variety of generative tasks, including interpolation, sample generation from noise, PCA interpolation in low dimension, and a downstream classification task.
Researcher Affiliation Industry Li Jing Facebook AI Research New York ljng@fb.com Jure Zbontar Facebook AI Research New York jzb@fb.com Yann Le Cun Facebook AI Research New York yann@fb.com
Pseudocode No The paper describes the method and architecture but does not include any pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statement about releasing source code or a link to a code repository.
Open Datasets Yes We train our model on the MNIST dataset [15] and the Celeb A dataset [16].
Dataset Splits No Each plot in Figures 2 and 4 depicts singular values (sorted from large to small) of the covariance matrix of the latent variables z corresponding to examples in the validation set. IRMAE l = 2 yields excellent reconstructions on validation set.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU, CPU models, memory) used for running the experiments.
Software Dependencies No Each model is trained with an Adam optimizer with a learning rate of 0.001.
Experiment Setup Yes We set the latent dimension to 128/512 for the two datasets, respectively. We use 8/4 extra linear matrices for regularization in IRMAE, respectively. ... Each model is trained with an Adam optimizer with a learning rate of 0.001. ... This MLP head has two linear layers of hidden dimension 128, with Re LU activation.