Inference-Path Optimization via Circuit Duplication in Frozen Visual Transformers for Marine Species Classification

TL;DR

This paper introduces a novel inference-time method called Circuit Duplication applied to frozen visual transformers, significantly improving marine species classification accuracy without model fine-tuning.

Contribution

It demonstrates the first application of Circuit Duplication in computer vision, enhancing frozen embeddings for better classification performance in marine species identification.

Findings

Circuit Duplication improves classification accuracy over standard frozen forward passes.

Class-specific circuit selection achieves near fully supervised performance, with some species exceeding it.

Approximately 75% of classes benefit from class-specific circuit duplication.

Abstract

Automated underwater species classification is constrained by annotation cost and environmental variation that limits the transferability of fully supervised models. Recent work has shown that frozen embeddings from self-supervised vision foundation models already provide a strong label-efficient baseline for marine image classification. Here we investigate whether this frozen-embedding regime can be improved at inference time, without fine-tuning or changing model weights. We apply Circuit Duplication, an inference-time method originally proposed for Large Language Models, in which a selected range of transformer layers is traversed twice during the forward pass. We evaluate on the class-imbalanced AQUA20 benchmark using frozen DINOv3 embeddings under two settings: global circuit selection, where a single duplicated circuit is chosen for the full dataset, and class-specific circuit selection, where each species may receive a different optimal circuit. Both settings use simple semi-supervised downstream classifiers. Circuit Duplication consistently improves over the standard frozen forward pass. At the maximum label budget, class-specific selection reaches a macro F1 of 0.875, closing the gap to the fully supervised ConvNeXt benchmark (0.889) to 1.4 points without any gradient-based training. Four species exceed their fully supervised reference, with octopus improving by +12.1 F1 points. Across all budgets, roughly 75% of classes prefer a class-specific circuit, indicating a genuinely class-dependent benefit. To our knowledge, this is the first application of Circuit Duplication to computer vision.