Deep Learning Alternatives of the Kolmogorov Superposition Theorem
For researchers in scientific computing and PDE simulation, ActNet offers a new KST-based architecture that improves upon prior KST-based networks, though its advantage over existing MLP methods is competitive rather than dominant.
The paper introduces ActNet, a scalable deep learning model based on the Kolmogorov Superposition Theorem, which outperforms Kolmogorov-Arnold Networks (KANs) across multiple benchmarks and is competitive with the best MLP-based approaches in Physics-Informed Neural Networks for solving partial differential equations.
This paper explores alternative formulations of the Kolmogorov Superposition Theorem (KST) as a foundation for neural network design. The original KST formulation, while mathematically elegant, presents practical challenges due to its limited insight into the structure of inner and outer functions and the large number of unknown variables it introduces. Kolmogorov-Arnold Networks (KANs) leverage KST for function approximation, but they have faced scrutiny due to mixed results compared to traditional multilayer perceptrons (MLPs) and practical limitations imposed by the original KST formulation. To address these issues, we introduce ActNet, a scalable deep learning model that builds on the KST and overcomes many of the drawbacks of Kolmogorov's original formulation. We evaluate ActNet in the context of Physics-Informed Neural Networks (PINNs), a framework well-suited for leveraging KST's strengths in low-dimensional function approximation, particularly for simulating partial differential equations (PDEs). In this challenging setting, where models must learn latent functions without direct measurements, ActNet consistently outperforms KANs across multiple benchmarks and is competitive against the current best MLP-based approaches. These results present ActNet as a promising new direction for KST-based deep learning applications, particularly in scientific computing and PDE simulation tasks.