Repurposing Obsolete Representations for Post-Deployment Adaptation

2026-10-01

AI Summary

This study addresses the challenge of partial output space invalidation in deployed deep neural networks, where fine-tuning incurs prohibitive costs. To this end, this work proposes Deep Repurposing, a novel framework that introduces a gradient-free analytical repair mapping mechanism. By leveraging latent geometry estimation, the method identifies and eliminates obsolete components while repurposing their valid structural representations to support retained tasks, thereby enabling efficient post-hoc adaptation. Experimental evaluations demonstrate that the proposed framework completely eradicates obsolete predictions, matching or surpassing existing baselines across standard benchmarks. Notably, it achieves up to a 60-fold acceleration in adaptation speed compared to conventional approaches, offering a highly efficient solution for maintaining deployed models under evolving task requirements.

Abstract

Deep neural networks are increasingly deployed in long-lived systems, where task requirements may change after training. In such settings, part of the original output space may become obsolete: a class, prediction region, or learned behaviour may no longer be valid. Existing approaches either leave the obsolete behaviour intact or require fine-tuning, which can be expensive. We propose Deep Repurposing (DR), a post-hoc framework for adapting models under task obsolescence. DR estimates the latent geometry of obsolete and retained regions, removes obsolete-supporting components, and reallocates retained-compatible evidence through an analytic repair map without gradient updates. This yields repaired predictions and representations in which obsolete regions no longer act as valid outputs, while useful obsolete structure can support the retained task. Across multiple task settings, DR removes obsolete behaviour while preserving retained utility. More importantly, across classification benchmarks, DR matches or exceeds competing unlearning and editing baselines in retained accuracy, eliminates obsolete predictions, and adapts up to $60\times$ faster than competing unlearning methods.

Problem

• post-deployment adaptation
• task obsolescence
• model editing
• machine unlearning
• deep neural networks

Innovation

• Deep Repurposing
• Post-Deployment Adaptation
• Machine Unlearning
• Analytic Repair Map
• Latent Geometry