An Autoencoder is a type of artificial neural network used for unsupervised learning. It is designed to learn efficient representations of data, typically for the purpose of dimensionality reduction or feature learning. The architecture of an autoencoder consists of an encoder and a decoder. The encoder compresses the input data into a lower-dimensional representation (encoding), and the decoder reconstructs the original data from this encoding. The objective is to minimize the difference between the input and the reconstructed output, encouraging the model to capture the most important features of the data.


Disentangled Wasserstein Autoencoder for T-Cell Receptor Engineering

In protein biophysics, the separation between the functionally important residues (forming the active site or binding surface) and those that create the overall structure (the fold) is a well-established and fundamental concept. Identifying and modifying those functional sites is critical for protein engineering but computationally nontrivial, and requires significant domain knowledge. To automate this process from a data-driven perspective, we propose a disentangled Wasserstein autoencoder with an auxiliary classifier, which isolates the function-related patterns from the rest with theoretical guarantees. This enables one-pass protein sequence editing and improves the understanding of the resulting sequences and editing actionsinvolved. To demonstrate its effectiveness, we apply it to T-cell receptors (TCRs), a well-studied structure-function case. We show that our method can be used to alterthe function of TCRs without changing the structural backbone, outperforming several competing methods in generation quality and efficiency, and requiring only 10% of the running time needed by baseline models. To our knowledge, this is the first approach that utilizes disentangled representations for TCR engineering.

Conditional Image-to-Video Generation with Latent Flow Diffusion Models

Conditional image-to-video (cI2V) generation aims to synthesize a new plausible video starting from an image (e.g., a person’s face) and a condition (e.g., an action class label like smile). The key challenge of the cI2V task lies in the simultaneous generation of realistic spatial appearance and temporal dynamics corresponding to the given image and condition. In this paper, we propose an approach for cI2V using novel latent flow diffusion models (LFDM) that synthesize an optical flow sequence in the latent space based on the given condition to warp the given image. Compared to previous direct-synthesis-based works, our proposed LFDM can better synthesize spatial details and temporal motion by fully utilizing the spatial content of the given image and warping it in the latent space according to the generated temporally-coherent flow. The training of LFDM consists of two separate stages: (1) an unsupervised learning stage to train a latent flow auto-encoder for spatial content generation, including a flow predictor to estimate latent flow between pairs of video frames, and (2) a conditional learning stage to train a 3D-UNet-based diffusion model (DM) for temporal latent flow generation. Unlike previous DMs operating in pixel space or latent feature space that couples spatial and temporal information, the DM in our LFDM only needs to learn a low-dimensional latent flow space for motion generation, thus being more computationally efficient. We conduct comprehensive experiments on multiple datasets, where LFDM consistently outperforms prior arts. Furthermore, we show that LFDM can be easily adapted to new domains by simply finetuning the image decoder. Our code is available at

Improving Disentangled Text Representation Learning with Information Theoretical Guidance

Learning disentangled representations of natural language is essential for many NLP tasks, e.g., conditional text generation, style transfer, personalized dialogue systems, etc. Similar problems have been studied extensively for other forms of data, such as images and videos. However, the discrete nature of natural language makes the disentangling of textual representations more challenging (e.g., the manipulation over the data space cannot be easily achieved). Inspired by information theory, we propose a novel method that effectively manifests disentangled representations of text, without any supervision on semantics. A new mutual information upper bound is derived and leveraged to measure dependence between style and content. By minimizing this upper bound, the proposed method induces style and content embeddings into two independent low-dimensional spaces. Experiments on both conditional text generation and text-style transfer demonstrate the high quality of our disentangled representation in terms of content and style preservation.

Learning Deep Network Representations with Adversarially Regularized Autoencoders

The problem of network representation learning, also known as network embedding, arises in many machine learning tasks assuming that there exist a small number of variabilities in the vertex representations which can capture the “semantics” of the original network structure. Most existing network embedding models, with shallow or deep architectures, learn vertex representations from the sampled vertex sequences such that the low-dimensional embeddings preserve the locality property and/or global reconstruction capability. The resultant representations, however, are difficult for model generalization due to the intrinsic sparsity of sampled sequences from the input network. As such, an ideal approach to address the problem is to generate vertex representations by learning a probability density function over the sampled sequences. However, in many cases, such a distribution in a low-dimensional manifold may not always have an analytic form. In this study, we propose to learn the network representations with adversarially regularized autoencoders (NetRA). NetRA learns smoothly regularized vertex representations that well capture the network structure through jointly considering both locality-preserving and global reconstruction constraints. The joint inference is encapsulated in a generative adversarial training process to circumvent the requirement of an explicit prior distribution, and thus obtains better generalization performance. We demonstrate empirically how well key properties of the network structure are captured and the effectiveness of NetRA on a variety of tasks, including network reconstruction, link prediction, and multi-label classification.