contrastive learning

Combination of deep neural network with attention mechanism enhances the explainability of protein contact prediction

This study introduces an attention-based convolutional neural network for protein contact prediction, showing improved interpretability and predictive performance on CASP13 targets.

Predicting Epigenomic Functions of Genetic Variants in the Context of Neurodevelopment via Deep Transfer Learning

The paper introduces MetaChrom, a deep transfer learning framework for predicting epigenomic functions of genetic variants in neurodevelopment, showing improved accuracy over existing methods.

Image denoising for fluorescence microscopy by self-supervised transfer learning

This work improves fluorescence microscopy image denoising by combining self-supervised and transfer learning to reduce the need for task-specific training data.

SAINT: automatic taxonomy embedding and categorization by Siamese triplet network

SAINT is an alignment-free sequence comparison method using a Siamese triplet network for efficient and accurate taxonomy embedding and categorization.

Unsupervised Neural Network Models of the Ventral Visual Stream

This paper shows that unsupervised neural networks can model primate ventral visual stream responses as well as supervised methods, with brain-like representations from limited data.

Cross-Modality Protein Embedding for Compound-Protein Affinity and Contact Prediction

This study predicts compound-protein affinity and contact using cross-modality protein embeddings, showing improved performance over single-modality approaches.

Bias invariant RNA-seq metadata annotation

This paper introduces a deep-learning based domain adaptation algorithm for RNA-seq metadata annotation, achieving higher accuracy than existing methods for tissue, sample source, and patient sex prediction.

Self-supervised retinal thickness prediction enables deep learning from unlabeled data to boost classification of diabetic retinopathy

The paper proposes a self-supervised method to predict retinal thickness from fundus images, enabling efficient diabetic retinopathy classification with fewer annotations.

Towards a Better Understanding of Reverse-Complement Equivariance for Deep Learning Models in Regulatory Genomics

This paper evaluates strategies for enforcing reverse-complement equivariance in deep learning models for regulatory genomics, finding post-hoc conjoined models perform best.

Counterfactual Hypothesis Testing of Tumor Microenvironment Scenarios Through Semantic Image Synthesis

The paper introduces a generative model to test hypotheses about cell-cell interactions in tumor microenvironments using synthetic multiplexed images.

Mustache: Multi-scale Detection of Chromatin Loops from Hi-C and Micro-C Maps using Scale-Space Representation

MO_SCPLOWUSTACHEC_SCPLOW detects chromatin loops from Hi-C and Micro-C data using scale-space theory, identifying more loops than HiCCUPS and showing strong agreement between datasets.

Statistical Agnostic Mapping: A Framework In Neuroimaging Based On Concentration Inequalities

This paper introduces a framework for neuroimaging using concentration inequalities to improve model validation and hypothesis testing with small sample sizes.

Deep neural net tracking of human pluripotent stem cells reveals intrinsic behaviors directing morphogenesis

The paper tracks human pluripotent stem cells using deep neural networks to understand intrinsic behaviors during morphogenesis.

A convolutional neural network for common coordinate registration of high-resolution histology images

This paper introduces a convolutional neural network for common coordinate registration of high-resolution histology images, showing improved performance over existing methods.

Phylogeographic model selection using convolutional neural networks

This study applies convolutional neural networks to phylogeographic model selection, achieving high accuracy in demographic model identification compared to approximate Bayesian computation.

Self-Supervised Contrastive Learning of Protein Representations By Mutual Information Maximization

CPCProt pretrains protein embeddings by telling a protein's own next fragment from fragments of random proteins, reportedly matching benchmarked self-supervised models with 2-10% of their parameters.

Ensemble learning for robust knee cartilage segmentation: data from the osteoarthritis initiative.

This paper evaluates an ensemble learning approach for automated knee cartilage segmentation using MR images from the Osteoarthritis Initiative, achieving high Dice coefficients and improved anatomical validity.

Segmentation boosting with compensation methods in optical coherence tomography angiography images

This study introduces a novel method to improve blood vessel detection in OCT angiography images by compensating for motion artifacts using a modified U-Net architecture.

Super-human cell death detection with biomarker-optimized neural networks

This paper introduces BO-CNN, a neural network that detects cell death in live microscopy with 96% accuracy, faster than previous methods and without requiring stains or dyes.

Topological Feature Extraction and Visualization of Whole Slide Images using Graph Neural Networks

The paper introduces a framework combining graph neural networks and topological data analysis to extract and visualize features from whole-slide images, demonstrating its application in colon cancer staging.

A new method for counting reproductive structures in digitized herbarium specimens using Mask R-CNN

This paper develops a Mask R-CNN model to automate counting reproductive structures in herbarium specimens, achieving high accuracy for buds and fruits but lower for flowers.

Dataset Augmentation Allows Deep Learning-Based Virtual Screening To Better Generalize To Unseen Target Classes, And Highlight Important Binding Interactions

This paper shows that dataset augmentation improves deep learning models for virtual screening by making them better at generalizing to new protein targets and highlighting important binding interactions.

A Large-Scale Internal Validation Study of Unsupervised Virtual Trichrome Staining Technologies on Non-alcoholic Steatohepatitis Liver Biopsies

This study validates the use of GANs for virtual trichrome staining of NASH liver biopsies, showing high correlation with real stains and clinical outcomes.

FUNCTIONAL NETWORK CONNECTIVITY (FNC)-BASED GENERATIVE ADVERSARIAL NETWORK (GAN) AND ITS APPLICATIONS IN CLASSIFICATION OF MENTAL DISORDERS

This paper introduces an FNC-based GAN for classifying mental disorders from fMRI data, achieving higher accuracy than existing methods in MDD and schizophrenia classification.

DeepGANnel: Synthesis of fully annotated single molecule patch-clamp data using generative adversarial networks

This paper develops DeepGANnel, a GAN-based method to generate synthetic single molecule patch-clamp data with labels, addressing the lack of large training datasets for automated analysis tools.

Coevolution-based prediction of protein-protein interactions in polyketide biosynthetic assembly lines

The paper introduces PKSpop, a coevolution-based method to predict protein-protein interactions in PKS assembly lines, achieving 80% accuracy in predicting protein order.

Predicting sites of epitranscriptome modifications using unsupervised representation learning based on generative adversarial networks

The paper introduces MR-GAN, a GAN-based model for predicting epitranscriptome modification sites using unsupervised learning, showing significant improvement over one-hot encoding and SRAMP.

Functional modules from variable genes: Leveraging percolation to analyze noisy, high-dimensional data

The paper introduces a density-based clustering method using percolation theory to identify functionally related gene modules from noisy, high-dimensional RNA sequencing data across different organisms.

Deep Metric Learning Network using Proxies for Chromosome Classification and Retrieval in Karyotyping Test

This paper introduces Proxy-ResNeXt-CBAM, a metric learning network for chromosome classification and retrieval in karyotyping, achieving high accuracy and effective clustering of embeddings.

Mapping Mesoscale Axonal Projections in the Mouse Brain Using A 3D Convolutional Network

TrailMap is a 3D convolutional network that maps and quantifies axonal projections in cleared mouse brains, validated across multiple neuronal populations.