VectorAdam for Rotation Equivariant Geometry Optimization
This addresses a specific deficiency in optimization algorithms for geometry problems, offering an incremental improvement for researchers and practitioners in machine learning and geometry processing.
The paper tackled the problem that Adam is not rotation-equivariant for vector-valued parameters, causing artifacts and biases, and proposed VectorAdam, a simple modification that resolves these issues while maintaining or improving convergence rates in machine learning and geometric optimization tasks.
The Adam optimization algorithm has proven remarkably effective for optimization problems across machine learning and even traditional tasks in geometry processing. At the same time, the development of equivariant methods, which preserve their output under the action of rotation or some other transformation, has proven to be important for geometry problems across these domains. In this work, we observe that Adam $-$ when treated as a function that maps initial conditions to optimized results $-$ is not rotation equivariant for vector-valued parameters due to per-coordinate moment updates. This leads to significant artifacts and biases in practice. We propose to resolve this deficiency with VectorAdam, a simple modification which makes Adam rotation-equivariant by accounting for the vector structure of optimization variables. We demonstrate this approach on problems in machine learning and traditional geometric optimization, showing that equivariant VectorAdam resolves the artifacts and biases of traditional Adam when applied to vector-valued data, with equivalent or even improved rates of convergence.