FiCABU: A Fisher-Based, Context-Adaptive Machine Unlearning Processor for Edge AI

TL;DR

FiCABU introduces a hardware-software co-designed, efficient machine unlearning method for edge AI that significantly reduces computation and energy use while maintaining accuracy, enabling practical privacy compliance at the edge.

Contribution

It presents FiCABU, a novel Fisher-based, context-adaptive unlearning processor integrated into a RISC-V edge AI chip, optimized for resource-constrained environments.

Findings

Achieves random-guess forget accuracy on benchmarks.

Reduces computation by up to 87.52% and energy to 0.13% of baseline.

Maintains retain accuracy comparable to retraining-free methods.

Abstract

Machine unlearning, driven by privacy regulations and the "right to be forgotten", is increasingly needed at the edge, yet server-centric or retraining-heavy methods are impractical under tight computation and energy budgets. We present FiCABU (Fisher-based Context-Adaptive Balanced Unlearning), a software-hardware co-design that brings unlearning to edge AI processors. FiCABU combines (i) Context-Adaptive Unlearning, which begins edits from back-end layers and halts once the target forgetting is reached, with (ii) Balanced Dampening, which scales dampening strength by depth to preserve retain accuracy. These methods are realized in a full RTL design of a RISC-V edge AI processor that integrates two lightweight IPs for Fisher estimation and dampening into a GEMM-centric streaming pipeline, validated on an FPGA prototype and synthesized in 45 nm for power analysis. Across CIFAR-20 and PinsFaceRecognition with ResNet-18 and ViT, FiCABU achieves random-guess forget accuracy while matching the retraining-free Selective Synaptic Dampening (SSD) baseline on retain accuracy, reducing computation by up to 87.52 percent (ResNet-18) and 71.03 percent (ViT). On the INT8 hardware prototype, FiCABU further improves retain preservation and reduces energy to 6.48 percent (CIFAR-20) and 0.13 percent (PinsFaceRecognition) of the SSD baseline. In sum, FiCABU demonstrates that back-end-first, depth-aware unlearning can be made both practical and efficient for resource-constrained edge AI devices.