AI Insight
This paper investigates when finite probe-based representations are sufficient for learning properties of neural networks. The authors establish theoretical results showing that probing intermediate hidden layer representations provides more informative signals than using only final network outputs. They introduce HIDDENPROBE, an architecture that leverages hidden layer probe responses, demonstrating state-of-the-art performance across benchmarks involving both multilayer perceptrons and Transformers.
Why it matters
Understanding neural network properties through probing has practical implications for interpretability, model selection, and meta-learning. This work provides theoretical grounding for probe-based methods and offers a practical technique that could improve how we analyze and work with trained neural networks across various machine learning applications.
Understand the Science
⚠️ Preprint – Noch nicht peer-reviewed
Dieser Artikel wurde noch nicht von unabhängigen Experten begutachtet. Die Ergebnisse sind vorläufig und sollten mit Vorsicht interpretiert werden.
Abstract: Learning properties of neural networks has recently attracted growing interest, with existing approaches operating either directly on network parameters or through probe-based representations of network behavior. While probing methods have shown strong empirical performance, their theoretical foundations remain limited. In this work, we study when finite probe-based representations are sufficient for learning neural functionals. We establish general identification and universality results for probing, and show that using intermediate hidden representations can provide significantly more informative representations than relying only on final outputs. Motivated by these results, we introduce HIDDENPROBE, a simple architecture for learning from hidden probe responses. Across a range of neural functional benchmarks, including both MLPs and Transformers, HIDDENPROBE consistently improves over existing probing methods and achieves state-of-the-art performance. Our code is publicly available on GitHub.
Source: Finite Probes Suffice: Identifiability and Universality for Weight-Space Learning