Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Ettori, Davide, Darabi, Nastaran, Senthilkumar, Sureshkumar, Trivedi, Amit Ranjan
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2509.15724
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917253356716032
author Ettori, Davide
Darabi, Nastaran
Senthilkumar, Sureshkumar
Trivedi, Amit Ranjan
author_facet Ettori, Davide
Darabi, Nastaran
Senthilkumar, Sureshkumar
Trivedi, Amit Ranjan
contents Large deep learning models such as BERT and ResNet achieve state-of-the-art performance but are costly to deploy at the edge due to their size and compute demands. We present RMT-KD, a compression method that leverages Random Matrix Theory (RMT) for knowledge distillation to iteratively reduce network size. Instead of pruning or heuristic rank selection, RMT-KD preserves only informative directions identified via the spectral properties of hidden representations. RMT-based causal reduction is applied layer by layer with self-distillation to maintain stability and accuracy. On GLUE and CIFAR-10, RMT-KD achieves up to 80% parameter reduction with only 2% accuracy loss, delivering 2.8x faster inference and nearly halved power consumption. These results establish RMT-KD as a mathematically grounded approach to network distillation.
format Preprint
id arxiv_https___arxiv_org_abs_2509_15724
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RMT-KD: Random Matrix Theoretic Causal Knowledge Distillation
Ettori, Davide
Darabi, Nastaran
Senthilkumar, Sureshkumar
Trivedi, Amit Ranjan
Machine Learning
Large deep learning models such as BERT and ResNet achieve state-of-the-art performance but are costly to deploy at the edge due to their size and compute demands. We present RMT-KD, a compression method that leverages Random Matrix Theory (RMT) for knowledge distillation to iteratively reduce network size. Instead of pruning or heuristic rank selection, RMT-KD preserves only informative directions identified via the spectral properties of hidden representations. RMT-based causal reduction is applied layer by layer with self-distillation to maintain stability and accuracy. On GLUE and CIFAR-10, RMT-KD achieves up to 80% parameter reduction with only 2% accuracy loss, delivering 2.8x faster inference and nearly halved power consumption. These results establish RMT-KD as a mathematically grounded approach to network distillation.
title RMT-KD: Random Matrix Theoretic Causal Knowledge Distillation
topic Machine Learning
url https://arxiv.org/abs/2509.15724