ProtoSeam: Lifting Classifier Training with Latent Gaussian Mixture Models

arXiv:2609.35174v1 Announce Type: new Abstract: We propose a lifted reformulation of supervised classification that improves the final accuracy of standard classifiers without changing the architecture at inference time. A network N=N_2irc N_1 is split at a single semantic interface and one learnable prototype per class is inserted there. Training combines a quadratic consensus penalty that pulls N_1(x) toward the prototype of its class with a classification loss of N_2 evaluated on samples drawn around the prototypes, whereat no gradient crosses the interface. At inference the prototypes are discarded and the unmodified network N_2irc N_1 is used. Across CIFAR-10, CIFAR-100, and TinyImageNet with ResNet and vision transformer backbones, lifted training improves test accuracy by up to five percentage points over variants without lifting under a shared tuning protocol. Moreover, we provide theoretical justification of those results.

Source: arXiv cs.LG | 2026-09-29