PPLNs: Parametric Piecewise Linear Networks for Event-Based Temporal Modeling and Beyond
This work addresses temporal modeling for event cameras and vision tasks, offering a novel method that is incremental in building on recent ideas like Kolmogorov-Arnold Networks.
The paper tackles temporal vision inference by introducing Parametric Piecewise Linear Networks (PPLNs), which model artificial neuron membrane potentials with learnable parametric functions, achieving state-of-the-art performance in event-based and image-based applications like steering prediction, human pose estimation, and motion deblurring.
We present Parametric Piecewise Linear Networks (PPLNs) for temporal vision inference. Motivated by the neuromorphic principles that regulate biological neural behaviors, PPLNs are ideal for processing data captured by event cameras, which are built to simulate neural activities in the human retina. We discuss how to represent the membrane potential of an artificial neuron by a parametric piecewise linear function with learnable coefficients. This design echoes the idea of building deep models from learnable parametric functions recently popularized by Kolmogorov-Arnold Networks (KANs). Experiments demonstrate the state-of-the-art performance of PPLNs in event-based and image-based vision applications, including steering prediction, human pose estimation, and motion deblurring. The source code of our implementation is available at https://github.com/chensong1995/PPLN.