SENTINEL Research Lab

Prof. Fekri is the founder of the SENTINEL Research Lab that has a multidisciplinary flavor in four intertwined broad areas of 1. Information, 2. Processing /Learning, 3. Communications, and 4. Decision Making. In particular, the team applies machine learning and logical reasoning, statistics and information theory to fundamental research problems in Science and Engineering.

Currently, the Research Lab has five main thrusts: (i) Semantic Communications in which the team develops the foundation of information theory for communication among distributed or federated rational decision makers. (ii) Learning via Logical Reasoning, where the team has focused on inductive/abductive/deductive reasoning over data or dynamical events. (iii) Reinforcement Learning RL in which the Lab investigates as to how incorporate neuro-symbolic logical reasoning into RL policy learning. (iv) Neuro-Logical Reasoning in Language Models, the research studies how to perform logical grounding of language models to generate reliable outputs. (v) Causal Discovery and Probabilistic Modeling, where the graphical model of the data (either in the form of causal graphs or Markov Random Fields) is learned by using (observational and possibly interventional) measurements from the variables of interest.

The team also is very active in Biomarker Sensing and the related Inverse Problems. The research investigates several problems including learning probabilistic graphical modeling of variables from measurement data, discovering the sensing mechanisms for the biomarkers, and developing theoretical foundations (using the density evolution principle) for optimum biomarker sensing for disease detection or therapy. Finally, the research Lab studies Molecular Communication, in which the team investigates the usage of molecules to encode and transmit information among nanomachines such as synthetic biological agents. The establishment of the foundations of molecular information theory, the development of the transceiver architecture and signaling techniques are been studied.

In the past, Prof. Fekri’s Lab investigated the theory and practice of error correction codes, the application of mathematical tools to modern networking, spanning from network packet compression and performance characterization of wired/wireless networks to the design, analysis, and optimization of communication protocols. The SENTINEL lab also developed novel frameworks for approximate computing, neural computing as well as social computing. The team also developed the first cryptographic system using finite field wavelet transforms. Follow the links in below for details of the current and past projects. You may also click on the "Project" tab at the top of this page

• Current Research Projects
  • Learning via Inductive Logic Reasoning
  • Semantic Communication for AI
  • Neuro-Symbolic Reinforcement Learning
  • Language Models and Neuro-Logic Reasoning
  • Causal Discovery from Data
  • Biomarker Sensing
  • Molecular Communication
• Past Research Projects
  • Coding Theory
  • Security
  • Network Information Theory
  • Network Compression
  • Neural Computing
  • Approximate Computing
  • Social Computing