Research from the Multicom Lab has made lasting contributions to information theory, data compression, multimedia communications, and machine learning, with impact spanning both academia and industry at significant scale.
Our work has directly shaped modern practice in image and video coding: the soft-decision quantization algorithms developed in the lab have been incorporated into the reference software of all video coding standards since H.264, and widely deployed across computing and communications devices — reaching billions of people worldwide. The lab’s compression technologies were also commercialized through SlipStream Data Inc., later acquired by BlackBerry.
More recently, the lab's information-theoretic approach to deep learning has produced results on model accuracy, adversarial robustness, knowledge distillation, and model intellectual property protection that have appeared at leading venues including IEEE Transactions on Information Theory, ICLR, NeurIPS, and ECCV. Members of the Multicom Research Group have also played leadership roles in the international research community, contributing to major conferences and journals in information theory, multimedia communications, and artificial intelligence.