Assistant Professor, University of Illinois at Chicago
Dr. Ravi's research involves developing machine learning models for computer vision tasks including algorithms for biomedical image analysis. In the past, he has invented novel numerical procedures to find optimal and reliable solutions for many problems such as hashing for compression, support vector machine based classification, and deep network training that are widely used in complex computer vision pipelines. More recently, he has been focusing on problems related to multimodal dataset pipelines where different modalities such as images, point cloud, text and so on are fused together for improved inference. Multimodal dataset analysis presents new challenges during training and inference such as different number of features, scale, dimensions, signal to noise ratio across modalities. To tackle such challenges, he is investigating applicability of implicit layers in a plug and play manner to train and predict efficiently in terms of memory and time.