Generative Adversarial Networks (GANs) have made releasing of synthetic images a viable approach to share data without releasing the original dataset. It has been shown that such synthetic data can be used for a variety of downstream tasks such as training classifiers that would otherwise require the original dataset to be shared. However, recent work has shown that the GAN models and their synthetically generated data can be used to infer the training set membership by an adversary who has access to the entire dataset and some auxiliary information. Current approaches to mitigate this problem (such as DPGAN) lead to dramatically poorer generated sample quality than the original non--private GANs. Here we develop a new GAN architecture (privGAN), where the generator is trained not only to cheat the discriminator but also to defend membership inference attacks. The new mechanism provides protection against this mode of attack while leading to negligible loss in downstream performances. In addition, our algorithm has been shown to explicitly prevent overfitting to the training set, which explains why our protection is so effective. The main contributions of this paper are: i) we propose a novel GAN architecture that can generate synthetic data in a privacy preserving manner without additional hyperparameter tuning and architecture selection, ii) we provide a theoretical understanding of the optimal solution of the privGAN loss function, iii) we demonstrate the effectiveness of our model against several white and black--box attacks on several benchmark datasets, iv) we demonstrate on three common benchmark datasets that synthetic images generated by privGAN lead to negligible loss in downstream performance when compared against non--private GANs.
github repo: https://github.com/microsoft/privGAN
UNCURL is s an unsupervised/semi-supervised pre-processing framework for scRNA-seq data. Given highly sampled/sparse transcriptomic data, UNCURL estimates the true state using a matrix factorization approach. This cleaned-up data can then be used in downstream applications such as visualizations, clustering and lineage estimation. UNCURL also allows users to provide qualitative prior information to semi-supervise the state estimation process. UNCURL is highly scalable and is shown to work on datasets with over 1 million cells.
Interactive app: https://uncurl.cs.washington.edu/
The goal of AccessMap is to enable safe, accessible trip planning on pedestrian ways for people with limited mobility. To meet that end, the people working on the AccessMap project develop tools for trip planning as well as tools for gathering and maintaining open data about sidewalks, curb ramps, construction information, etc.
Multi-view evidence aggregation
Late onset Alzheimer’s disease (LOAD) is currently a disease with no known effective treatment options. To address this, there have been a recent surge in the generation of multi-modality data (Hodes and Buckholtz, 2016; Muelleret al., 2005) to understand the biology of the disease and potential drivers that causally regulate it. However, most analytic studies using these data-sets focus on uni-modal analysis of the data. Here we propose a data-driven approach to integrate multiple data types and analytic outcomes to aggregate evidences to support the hypothesis that a gene is a genetic driver of the disease. The main algorithmic contributions of our paper are: i) A general machine learning framework to learn the key characteristics of a few known driver genes from multiple feature-sets and identifying other potential driver genes which have similar feature representations, and ii) A flexible ranking scheme with the ability to integrate external validation in the form of Genome Wide Association Study (GWAS) summary statistics. While we currently focus on demonstrating the effectiveness of the approach using different analytic outcomes from RNA-Seq studies, this method is easily generalizable to other data modalities and analysis types.
We demonstrate the utility of our machine learning algorithm on two benchmark multi-view datasets by significantly outperforming the baseline approaches in predicting missing labels. We then use the algorithm to predict and rank potential drivers of Alzheimers. We show that our ranked genes show a significant enrichment for SNPs associated with Alzheimers, and are enriched in pathways that have been previously associated with the disease.
Source code and link to all feature sets is availabile at https://github.com/Sage-Bionetworks/EvidenceAggregatedDriverRanking.