AI RESEARCH
HalfNet: Randomized Neural Networks with Learned Subspace Geometry
arXiv CS.LG
•
ArXi:2606.04583v1 Announce Type: new Many researchers investigated neural networks with some of their weights fixed to values randomly drawn from a given distribution, e.g., $N(0, I)$. Our proposed HalfNet draws random weights from $N(0, \Sigma)$, where $\Sigma$, which defines the geometry of the distribution, has a low-rank factorization that we learn from data. Experiments on MNIST and CIFAR-10 nstrate that HalfNet can match the performance of fully trained multilayer perceptrons while using substantially fewer parameters.