The Deep Neural Network (DNN) has significantly enhanced the performance of pattern classification and time series predictions for numerous important real-world applications. However, the complexity of a typical DNN network has also made it difficult to incorporate such an algorithm in low power mobile devices and internet of things. In this work, we present methods to approximate the performance of a trained DNN by modifying its structural design and discuss specific design cases.

Social interactions provide support for individuals and heavily influence the decisions they make. Understanding how information is extracted from social interactions and how that information influences our decisions could have a tremendous public health benefit. In this talk, I will describe experiments designed to elucidate how the brain constructs representations of complex social interactions and the network modeling and machine learning methods developed to study those processes.

In many fields today, multiple sets of data are readily available. These might either be multimodal data where information about a given phenomenon is obtained through different types of acquisition techniques resulting in datasets with complementary information but essentially of different types, or multiset data where the datasets are all of the same type but acquired from different samples, at different time points, or under different conditions.