A5047050773- Single-cell and spatial transcriptomics
- Gene Regulatory Network Analysis
- Gene expression and cancer classification
- Stochastic Gradient Optimization Techniques
- Neural Networks and Applications
- Immune cells in cancer
- Machine Learning and Algorithms
- Advanced Neural Network Applications
- Advanced SAR Imaging Techniques
- Adversarial Robustness in Machine Learning
- Extracellular vesicles in disease
- Gaussian Processes and Bayesian Inference
- Domain Adaptation and Few-Shot Learning
Stanford University
2016-2020
Stanford Medicine
2017-2018
Current approaches to single-cell transcriptomic analysis are computationally intensive and require assay-specific modeling, which limits their scope generality. We propose a novel method that compares clusters cells based on transcript-compatibility read counts rather than the transcript or gene quantifications used in standard pipelines. In reanalysis of two landmark yet disparate RNA-seq datasets, we show our is up orders magnitude faster previous approaches, provides accurate some cases...
With the recent proliferation of single-cell RNA-Seq experiments, several methods have been developed for unsupervised analysis resulting datasets. These often rely on unintuitive hyperparameters and do not explicitly address subjectivity associated with clustering.
Deep neural networks (DNNs) have set benchmarks on a wide array of supervised learning tasks. Trained DNNs, however, often lack robustness to minor adversarial perturbations the input, which undermines their true practicality. Recent works increased DNNs by fitting using adversarially-perturbed training samples, but improved performance can still be far below seen in non-adversarial settings. A significant portion this gap attributed decrease generalization due training. In work, we extend...
Summary Single-cell computational pipelines involve two critical steps: organizing cells (clustering) and identifying the markers driving this organization (differential expression analysis). State-of-the-art perform differential analysis after clustering on same dataset. We observe that because forces separation, reusing dataset generates artificially low p -values hence false discoveries. introduce a valid post-clustering framework which corrects for problem. provide software at...
The success of deep neural networks stems from their ability to generalize well on real data; however, et al. have observed that can easily overfit randomly-generated labels. This observation highlights the following question: why do gradient methods succeed in finding generalizable solutions for while there exist with poor generalization behavior? In this work, we use a Fourier-based approach study properties gradient-based over 2-layer band-limited activation functions. Our results...
ABSTRACT Background With the recent proliferation of single-cell RNA-Seq experiments, several methods have been developed for unsupervised analysis resulting datasets. These often rely on unintuitive hyperparameters and do not explicitly address subjectivity associated with clustering. Results In this work, we present DendroSplit, an interpretable framework analyzing datasets that addresses both clustering interpretability issues. DendroSplit offers a novel perspective problem motivated by...