Energy-hungry neural nets move slowly. That’s why Team LATTICE built a Ferroelectric Diode crossbar array capable of compute-in memory operations with an FPF neural network capable of recognizing handwritten digits from the MNIST dataset. Because neural networks are primarily driven by I/O memory costs, the general time complexity of matrix multiplication leads to wider problems. As Generative AI grows, so does the usage of its power-hungry chips.
Problem Impact: Time and energy-efficient, LATTICE is a heat-resistant matrix multiplication device. The impact of this solution will be significant on not only current leading Generative AI companies, but also on edge AI applications like agriculture and space. This solution will also be important in the public sector, as many energy grids are unable to sustain the large energy demands from current data centers.
Team Members: Zirun Han, Alexander Kyimpopkin, Rose Wang, Spencer Ware
Faculty Advisors: Troy Olsson and Deep Jariwala