The Fast Machine Learning for Science Conference was held at UCSD in early September. Our group had a strong showing. Ray presented a poster on our FlowFI project, which assesses machine learning accelerators’ resilience to faults. Andy gave a talk about SliceFI, a fault-injection tool that enables efficient NN robustness analysis by requiring fewer model computations per fault injection. Zhenghua gave a talk on “Design Rules for Extreme-Edge Scientific Computing on AI Engines.” Liv gave the conference summary where she summarized the talks and posters and talked about how she “grew up” in this research community. The FlowFI project won a “Best Poster Award,” and the Design Rules project was given a “Best Talk Award.” A tremendous showing from our research group!



