A5029376641- Neural Networks and Applications
- Advanced Neural Network Applications
- Multimodal Machine Learning Applications
- Domain Adaptation and Few-Shot Learning
- RNA Research and Splicing
- RNA modifications and cancer
- Trace Elements in Health
- Generative Adversarial Networks and Image Synthesis
- Neural Networks and Reservoir Computing
- Reinforcement Learning in Robotics
- Stochastic Gradient Optimization Techniques
- RNA regulation and disease
- Advanced Image and Video Retrieval Techniques
Ontario Genomics
2023
Birla Institute of Technology and Science - Hyderabad Campus
2018
Abstract Accurately modeling and predicting RNA biology has been a long-standing challenge, bearing significant clinical ramifications for variant interpretation the formulation of tailored therapeutics. We describe foundation model biology, “BigRNA”, which was trained on thousands genome-matched datasets to predict tissue-specific expression, splicing, microRNA sites, binding protein specificity from DNA sequence. Unlike approaches that are restricted missense variants, BigRNA can identify...
Recurrent neural networks are known for their notorious exploding and vanishing gradient problem (EVGP). This becomes more evident in tasks where the information needed to correctly solve them exist over long time scales, because EVGP prevents important components from being back-propagated adequately a large number of steps. We introduce simple stochastic algorithm (\textit{h}-detach) that is specific LSTM optimization targeted towards addressing this problem. Specifically, we show when...
Attention and self-attention mechanisms, are now central to state-of-the-art deep learning on sequential tasks. However, most recent progress hinges heuristic approaches with limited understanding of attention's role in model optimization computation, rely considerable memory computational resources that scale poorly. In this work, we present a formal analysis how affects gradient propagation recurrent networks, prove it mitigates the problem vanishing gradients when trying capture long-term...
Growth of multimodal content on the web and social media has generated abundant weakly aligned image-sentence pairs. However, it is hard to interpret them directly due intrinsic intension. In this paper, we aim annotate such pairs with connotations as labels capture We achieve a connotation embedding model (CMEM) using novel loss function. It's unique characteristics over previous models include: (i) exploitation data opposed only visual information, (ii) robustness outlier in multi-label...
<title>Abstract</title> Accurately modeling and predicting RNA biology has been a long-standing challenge, bearing significant clinical ramifications for variant interpretation the formulation of tailored therapeutics. We describe foundation model biology, “BigRNA”, which was trained on thousands genome-matched datasets to predict tissue-specific expression, splicing, microRNA sites, binding protein specificity from DNA sequence. Unlike approaches that are restricted missense variants,...