10 Times Faster, 10 Times Less Energy: Solving AI's Memory Bottleneck With Algorithms and Coding Theory - USC Viterbi School of Engineering
9/10USC Viterbi School of Engineering unveiled new algorithms based on coding theory on May 4, 2026, that enhance AI model training speed by 10 times while decreasing energy use by a factor of 10, addressing critical memory bottlenecks in AI systems.
