Major Research Areas
What is the basis of intelligence? How can we uncover this by directly interacting with the human brain? Is it possible to engineer intelligence into machines? How can we utilize advances in artificial intelligence to improve patient care? These are among the questions that inspire us at AISINAI. The following four areas represent our major, active research verticals that capture our attempts to answer these questions. While we are always trying to find new problems and new solutions, the below areas are our major ones:
Machine Learning and deep learning
Fundamental investigations into machine learning and deep learning have the potential to transform everything else that we do. We are guided to areas that are of specific relevance to the biomedical space which is characterized by expensive data, and sparse labels: weakly supervised learning, reinforcement learning with imperfect information, and transfer learning. We are particularly interested in finding ways to utilize our unique access to the human brain to drive fundamental discoveries in deep learning.
basic and clinical Neurosciences
Neuroscience and biological learning present are a major area of interest to us for two reasons. First, many neuroscience generate significant quantities of data that can benefit from machine learning techniques for analysis. Second, we believe that by advancing our understanding of the human brain and biological learning that we will derive new insights into machine learning and artificial intelligence. By being embedded within the Mount Sinai Health System and closely affiliated with the Department of Neurological Surgery, we have a unique and direct access to the human brain to drive our research into understanding the basis of intelligence.
Radiology and biomedical imaging
Machine learning based approaches to studying and improving radiological and biomedical imaging have the promise to radically change the practice of clinical medicine and basic scientific research. We study both discriminative and generative models to attempt to improve clinical outcomes, and accelerate scientific research.
Natural language processing and clinical projects
Natural language processing is often an essential step towards obtaining features for studying clinical problems in medicine. Multiple projects investigate both natural language processing techniques themselves, their application to clinical text, and the subsequent utilization of clinical text in a variety of projects spanning the inpatient ICUs to the outpatient wards and even population health.