Learnability Window in Gated Recurrent Neural Networks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Author: Livi, Lorenzo
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915895954112512
author Livi, Lorenzo
author_facet Livi, Lorenzo
contents We develop a statistical theory of temporal learnability in recurrent neural networks, quantifying the maximal temporal horizon $\mathcal{H}_N$ over which gradient-based learning can recover lag-dependent structure at finite sample size $N$. The theory is built on the effective learning rate envelope $f(\ell)$, a functional that captures how gating mechanisms and adaptive optimizers jointly shape the coupling between state-space transport and parameter updates during Backpropagation Through Time. Under heavy-tailed ($α$-stable) fluctuations, where empirical averages concentrate at rate $N^{-1/κ_α}$ with $κ_α= α/(α-1)$, the interplay between envelope decay and statistical concentration yields explicit scaling laws for the growth of $\mathcal{H}_N$: logarithmic, polynomial, and exponential temporal learning regimes emerge according to the decay law of $f(\ell)$. These results identify the envelope decay as the key determinant of temporal learnability: slower attenuation of $f(\ell)$ enlarges the learnability window $\mathcal{H}_N$, while heavy-tailed noise compresses temporal horizons by weakening statistical concentration. Experiments across multiple gated architectures and optimizers corroborate these structural predictions.
format Preprint
id arxiv_https___arxiv_org_abs_2512_05790
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learnability Window in Gated Recurrent Neural Networks
Livi, Lorenzo
Machine Learning
Data Analysis, Statistics and Probability
We develop a statistical theory of temporal learnability in recurrent neural networks, quantifying the maximal temporal horizon $\mathcal{H}_N$ over which gradient-based learning can recover lag-dependent structure at finite sample size $N$. The theory is built on the effective learning rate envelope $f(\ell)$, a functional that captures how gating mechanisms and adaptive optimizers jointly shape the coupling between state-space transport and parameter updates during Backpropagation Through Time. Under heavy-tailed ($α$-stable) fluctuations, where empirical averages concentrate at rate $N^{-1/κ_α}$ with $κ_α= α/(α-1)$, the interplay between envelope decay and statistical concentration yields explicit scaling laws for the growth of $\mathcal{H}_N$: logarithmic, polynomial, and exponential temporal learning regimes emerge according to the decay law of $f(\ell)$. These results identify the envelope decay as the key determinant of temporal learnability: slower attenuation of $f(\ell)$ enlarges the learnability window $\mathcal{H}_N$, while heavy-tailed noise compresses temporal horizons by weakening statistical concentration. Experiments across multiple gated architectures and optimizers corroborate these structural predictions.
title Learnability Window in Gated Recurrent Neural Networks
topic Machine Learning
Data Analysis, Statistics and Probability
url https://arxiv.org/abs/2512.05790