Efficient Learning of Generative Models via Finite-Difference Score Matching
This work addresses a computational bottleneck for researchers and practitioners in generative modeling, offering an incremental improvement in efficiency.
The paper tackles the computational inefficiency of optimizing higher-order derivatives in generative modeling, specifically for score matching, by introducing a finite-difference approximation method that reduces cost and improves stability while achieving comparable results.
Several machine learning applications involve the optimization of higher-order derivatives (e.g., gradients of gradients) during training, which can be expensive in respect to memory and computation even with automatic differentiation. As a typical example in generative modeling, score matching (SM) involves the optimization of the trace of a Hessian. To improve computing efficiency, we rewrite the SM objective and its variants in terms of directional derivatives, and present a generic strategy to efficiently approximate any-order directional derivative with finite difference (FD). Our approximation only involves function evaluations, which can be executed in parallel, and no gradient computations. Thus, it reduces the total computational cost while also improving numerical stability. We provide two instantiations by reformulating variants of SM objectives into the FD forms. Empirically, we demonstrate that our methods produce results comparable to the gradient-based counterparts while being much more computationally efficient.