Abstract

The goal of this project is to leverage ionic computing as a highly promising and novel opportunity to enable the needed, revolutionary improvements in computing energy efficiency. If current trends continue, the global power requirements for computing will reach global primary power production capacity by 2040. The goal for the field at present is to improve the energy efficiency of computing by more than a million fold. This project will advance a novel ionic computing device (sketched below), that we call Electrochemical Ionic Synapse (EIS), that is inspired from how our biological synapses function. In order to reduce the operating voltage while operating at nano-second speed regime, we will explore promising materials based on CeO2. Use of a good proton conductor (nano-porous CeO2 electrolyte), together with a doped CeO2 channel (Pr-doped CeO2 in particular) will enable a highquality interface with low resistance to proton transfer. High proton conductivity of the electrolyte and low interface resistance promise to improve the energy efficiency as well as reduce the operating voltage and improve the endurance of the EIS devices. Ultimately, this project will advance a novel device technology, to reduce the energy consumption and CO2 emissions of computing, while advancing the abilities of artificial intelligence hardware.

MIT PI
Bilge Yildiz, Professor, Departments of Nuclear Science and Engineering

PT PI
José P. B. Silva, Assistant Researcher at CF-UM-UP, Portugal

Abstract

This project advances the frontiers of eXplainable Artificial Intelligence (XAI). Modern high-performance prediction models (e.g., random forests, gradient boosting, neural networks, LLMs) are complex black-box objects that are difficult to interpret and use in critical applications. One promising XAI approach is rule-based distillation where a large ensemble of trees is replaced by a small number of rules with model utility almost comparable to the original complex model. Despite their promise, the current scope of these approaches is limited. Our goal is to advance the frontiers of rule-based distillation. Our proposed framework (i) significantly generalizes the notion of rules beyond their traditional usage; (ii) allows the distilled rule ensemble to incorporate various user-defined notions of interpretability, trust, transparency, etc while maximally retaining model utility. We propose integrating these interpretable compact models into performant LLMs for improved in-context learning.

MIT PI
Rahul Mazumder, Associate Professor, Sloan School of Management

PT PI
Paulo Cortez, Full Professor, Dep. of Information Systems (DSI), U. Minho

Abstract

Bispecific antibody drugs provide transformative cures, but they are often refractory to modern bioproduction methods. Here, we will apply large-scale data and AI to integrate manufacturing with early bispecific antibody discovery. We will collect high-throughput manufacturability data and train an AI-based drug design algorithm to enhance drug activity, potency, and product quality. First, we will clone large libraries of bispecific antibody variants into manufacture-ready cell lines and growth conditions. Next, each bispecific antibody-producing cell will be captured emulsion microreactor droplets to analyze functional performance in manufacturing-like conditions. High-throughput sequencing will be used to analyze test results en masse, and AI models will be trained to identify critical features of manufacturable molecules. Finally, we will apply trained AI algorithms to generate new manufacturing-ready bispecific designs and evaluate their improvement related to controls. If successful, we will establish a seamless transition between early discovery and large-scale manufacturing for bispecific antibody drug products.

MIT PI
Brandon DeKosky, Associate Professor, Department of Chemical Engineering

PT PI
Paula Alves, CEO of iBET, Professor at NOVA University of Lisbon, Portugal
Antonio Roldao, Head of Cell-based Vaccines Development Lab, Coordinator of Late-stage R&D and Bioproduction Unit, iBET, Portugal
Jose Escandell, Principal Scientist, Animal Cell Technology Unit iBET, Portugal
Patrícia Alves, Coordinator of Analytical Services Unit, iBET, Portugal