A Computing-in-Memory-Based One-Class Hyperdimensional Computing Model for Outlier Detection

In this work, we present <bold>ODHD</bold>, an algorithm for outlier detection based on hyperdimensional computing (HDC), a non-classical learning paradigm. Along with the HDC-based algorithm, we propose <bold>IM-ODHD</bold>, a computing-in-memory (CiM) implementation based on hardware/software (HW/SW) codesign for improved latency and energy efficiency. The training and testing phases of <bold>ODHD</bold> may be performed with conventional CPU/GPU hardware or our <bold>IM-ODHD</bold>, SRAM-based CiM architecture using the proposed HW/SW codesign techniques. We evaluate the performance of <bold>ODHD</bold> on six datasets from different application domains using three metrics, namely accuracy, F1 score, and ROC-AUC, and compare it with multiple baseline methods such as OCSVM, isolation forest, and autoencoder. The experimental results indicate that <bold>ODHD</bold> outperforms all the baseline methods in terms of these three metrics on every dataset for both CPU/GPU and CiM implementations. Furthermore, we perform an extensive design space exploration to demonstrate the tradeoff between delay, energy efficiency, and performance of <bold>ODHD</bold>. We demonstrate that the HW/SW codesign implementation of the outlier detection on <bold>IM-ODHD</bold> is able to outperform the GPU-based implementation of <bold>ODHD</bold> by at least 331.5<inline-formula><tex-math notation="LaTeX">$\times$</tex-math><alternatives><mml:math><mml:mo>×</mml:mo></mml:math><inline-graphic xlink:href="reis-ieq1-3371782.gif"/></alternatives></inline-formula>/889<inline-formula><tex-math notation="LaTeX">$\times$</tex-math><alternatives><mml:math><mml:mo>×</mml:mo></mml:math><inline-graphic xlink:href="reis-ieq2-3371782.gif"/></alternatives></inline-formula> in terms of training/testing latency (and on average 14.0<inline-formula><tex-math notation="LaTeX">$\times$</tex-math><alternatives><mml:math><mml:mo>×</mml:mo></mml:math><inline-graphic xlink:href="reis-ieq3-3371782.gif"/></alternatives></inline-formula>/36.9<inline-formula><tex-math notation="LaTeX">$\times$</tex-math><alternatives><mml:math><mml:mo>×</mml:mo></mml:math><inline-graphic xlink:href="reis-ieq4-3371782.gif"/></alternatives></inline-formula> in terms of training/testing energy consumption).