Ultrafast On-chip Online Learning via Spline Locality in Kolmogorov-Arnold Networks
This addresses the need for low-latency adaptation in critical systems, representing an incremental advance in applying KANs to resource-constrained hardware.
The paper tackled the problem of enabling ultrafast online learning for high-frequency systems like quantum computing controls by leveraging Kolmogorov-Arnold Networks (KANs) to achieve sub-microsecond latencies, demonstrating superior efficiency and expressiveness over MLPs on FPGAs.
Ultrafast online learning is essential for high-frequency systems, such as controls for quantum computing and nuclear fusion, where adaptation must occur on sub-microsecond timescales. Meeting these requirements demands low-latency, fixed-precision computation under strict memory constraints, a regime in which conventional Multi-Layer Perceptrons (MLPs) are both inefficient and numerically unstable. We identify key properties of Kolmogorov-Arnold Networks (KANs) that align with these constraints. Specifically, we show that: (i) KAN updates exploiting B-spline locality are sparse, enabling superior on-chip resource scaling, and (ii) KANs are inherently robust to fixed-point quantization. By implementing fixed-point online training on Field-Programmable Gate Arrays (FPGAs), a representative platform for on-chip computation, we demonstrate that KAN-based online learners are significantly more efficient and expressive than MLPs across a range of low-latency and resource-constrained tasks. To our knowledge, this work is the first to demonstrate model-free online learning at sub-microsecond latencies.