Sequencing the Neurome: Towards Scalable Exact Parameter Reconstruction of Black-Box Neural Networks
This enables scalable exact parameter reconstruction for security, verification, and interpretability, representing a major advance beyond existing methods.
The paper tackles the NP-hard problem of exactly reconstructing neural network parameters from query access, achieving reconstruction of networks with over 1.5 million parameters and 7 layers deep with max parameter difference less than 0.0001.
Inferring the exact parameters of a neural network with only query access is an NP-Hard problem, with few practical existing algorithms. Solutions would have major implications for security, verification, interpretability, and understanding biological networks. The key challenges are the massive parameter space, and complex non-linear relationships between neurons. We resolve these challenges using two insights. First, we observe that almost all networks used in practice are produced by random initialization and first order optimization, an inductive bias that drastically reduces the practical parameter space. Second, we present a novel query generation algorithm that produces maximally informative samples, letting us untangle the non-linear relationships efficiently. We demonstrate reconstruction of a hidden network containing over 1.5 million parameters, and of one 7 layers deep, the largest and deepest reconstructions to date, with max parameter difference less than 0.0001, and illustrate robustness and scalability across a variety of architectures, datasets, and training procedures.