LAIR-Net: Leaky Alignment-Impulse Residual Networks for Tabular Regression

Key points

• Deep randomized models fix hidden-layer parameters through random initialization and learn only closed-form readouts, typically adding depth by stacking random trans formations without target-aware control of hidden-state evolution.1
• We derive a depth uniform bound on input-perturbation sensitivity and use controlled simulations to attribute gains over a randomized baseline to the anchor rather than recursion or added capacity.1
• Benefits emerge when a nonlinear target structure is learnable at the available noise level and diminish for nearly linear targets or dominant noise.1
• Across 23 benchmark datasets, LAIR-Net achieves the best average rank among eight randomized networks and twelve conventional models, with relative performance associated with the same nonlinear-structure and noise quantities identified in simulation.1

Sources (1)

• [1]LAIR-Net: Leaky Alignment-Impulse Residual Networks for Tabular Regression
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 09:08 AM

  We propose LAIR Net, the Leaky Alignment-Impulse Residual Network, which mixes a shallow learned anchor into each hidden state through a leaky residual transition.

  Deep randomized models fix hidden-layer parameters through random initialization and learn only closed-form readouts, typically adding depth by stacking random trans formations without target-aware control of hidden-state evolution.

Extractive summary: sentences quoted from the sources.