Skip to search formSkip to main contentSkip to account menu

Attend and Predict: Understanding Gene Regulation by Selective Attention on Chromatin

An attention-based deep learning approach that uses a unified architecture to model and to interpret dependencies among chromatin factors for controlling gene regulation, and its attention scores provide a better interpretation than state-of-the-art feature visualization methods such as saliency maps.
View on PubMed (opens in a new tab)

Neural Message Passing for Multi-Label Classification

The proposed Label Message Passing (LaMP) Neural Networks to efficiently model the joint prediction of multiple labels are simple, accurate, interpretable, structure-agnostic, and applicable for predicting dense labels since LaMP is incredibly parallelizable.
View PDF on arXiv (opens in a new tab)

Perturbing Inputs for Fragile Interpretations in Deep Natural Language Processing

This paper demonstrates how interpretations can be manipulated by making simple word perturbations on an input text to attack two SOTA interpretation methods, across three popular Transformer models and on three different NLP datasets.
View on ACL (opens in a new tab)

DeepDiff: DEEP‐learning for predicting DIFFerential gene expression from histone modifications

A novel attention‐based deep learning architecture, DeepDiff, is developed that provides a unified and end‐to‐end solution to model and to interpret how dependencies among histone modifications control the differential patterns of gene regulation.
View on Wolters Kluwer (opens in a new tab)

GaKCo: a Fast Gapped k-mer string Kernel using Counting

This work proposes a fast algorithm for calculating Gapped k-mer Kernel using Counting (GaKCo), which uses associative arrays to calculate the co-occurrence of substrings using cumulative counting and provides a rigorous asymptotic analysis that compares GaKCo with the state-of-the-art gk-SK.
View PDF on arXiv (opens in a new tab)

Evolving Image Compositions for Feature Representation Learning

This paper proposes PatchMix, a data augmentation method that creates new samples by composing patches from pairs of images in a grid-like pattern that outperforms a base model on CIFAR-10, CIFar-100, Tiny Imagenet, and ImageNet and explores evolutionary search as a guiding strategy to jointly discover optimal grid- like patterns and image pairings.
View PDF on arXiv (opens in a new tab)

Transfer Learning for Predicting Virus-Host Protein Interactions for Novel Virus Sequences

The proposed DeepVHPPI, a novel deep learning framework combining a self-attention-based transformer architecture and a transfer learning training strategy to predict interactions between human proteins and virus proteins that have novel sequence patterns, outperforms the state-of-the-art methods significantly in predicting Virus–Human protein interactions.
View on ACM (opens in a new tab)

Fast and Scalable Learning of Sparse Changes in High-Dimensional Gaussian Graphical Model Structure

A novel method, DIFFEE is proposed for estimating DIFFerential networks via an Elementary Estimator under a high-dimensional situation and surprisingly achieves the same asymptotic convergence rates as the state-of-the-art estimators that are much more difficult to compute.
View PDF on arXiv (opens in a new tab)

A Fast and Scalable Joint Estimator for Integrating Additional Knowledge in Learning Multiple Related Sparse Gaussian Graphical Models

A novel hybrid norm is designed as the minimization objective to enforce the superposition of two weighted sparsity constraints, one on the shared interactions and the other on the task-specific structural patterns, which enables JEEK to elegantly consider various forms of existing knowledge based on the domain at hand and avoid the need to design knowledge-specific optimization.
View PDF on arXiv (opens in a new tab)

Transfer Learning with MotifTransformers for Predicting Protein-Protein Interactions Between a Novel Virus and Humans

This work proposes a novel deep learning architecture designed for in silico PPI prediction and a transfer learning approach to predict interactions between novel virus proteins and human proteins and shows that it outperforms the state-of-the-art methods significantly in predicting Virus-Human protein interactions for SARS-CoV-2, H1N1, and Ebola.
View PDF (opens in a new tab)
By clicking accept or continuing to use the site, you agree to the terms outlined in our Privacy Policy (opens in a new tab), Terms of Service (opens in a new tab), and Dataset License (opens in a new tab)