paper-with-me

홈 › Papers

Knowledge Distillation via Route Constrained Optimization

2019-04-19 · ICCV 2019 10 · Xiao Jin, Baoyun Peng, Yi-Chao Wu, Yu Liu, Jiaheng Liu, Ding Liang, Xiaolin Hu

Distillation-based learning boosts the performance of the miniaturized neural network based on the hypothesis that the representation of a teacher model can be used as structured and relatively weak supervision, and thus would be easily learned by a miniaturized model. However, we find that the representation of a converged heavy model is still a strong constraint for training a small student model, which leads to a high lower bound of congruence loss. In this work, inspired by curriculum learning we consider the knowledge distillation from the perspective of curriculum learning by routing. Instead of supervising the student model with a converged teacher model, we supervised it with some anchor points selected from the route in parameter space that the teacher model passed by, as we called route constrained optimization (RCO). We experimentally demonstrate this simple operation greatly reduces the lower bound of congruence loss for knowledge distillation, hint and mimicking learning. On close-set classification tasks like CIFAR100 and ImageNet, RCO improves knowledge distillation by 2.14% and 1.5% respectively. For the sake of evaluating the generalization, we also test RCO on the open-set face recognition task MegaFace.

📄 PDF Abstract BibTeX arXiv:1904.09149

Code (1)

SforAiDl/KD_Lib pytorch

Tasks

Face RecognitionKnowledge Distillation

Methods 이 논문이 사용한 방법론

Knowledge Distillation A very simple way to improve the performance of almost any machine learning algorithm is to train many different models on the same data and then to average their predictions.…

Similar Papers 제목 키워드 기반

RouteNLP: Closed-Loop LLM Routing with Conformal Cascading and Distillation Co-Optimization

2026-04-26 · Dongxin Guo, Jikun Wu, Siu Ming Yiu arxiv

Serving diverse NLP workloads with large language models is costly: at one enterprise partner, inference costs exceeded $200K/month despite over 70% of queries being routine tasks well within the capability of smaller mo…

Knowledge Distillation

Every Expert Matters: Towards Effective Knowledge Distillation for Mixture-of-Experts Language Models

2025-02-18 · Gyeongman Kim, Gyouk Chu, Eunho Yang

With the emergence of Mixture-of-Experts (MoE), the efficient scaling of model size has accelerated the development of large language models in recent years. However, their high memory requirements prevent their use in r…

Knowledge DistillationMixture-of-ExpertsModel Compression

DLink: Distilling Layer-wise and Dominant Knowledge from EEG Foundation Models

2026-04-16 · Jingyuan Wang, Zhihao Jia, Chenyu Liu, Xinliang Zhou 외 arxiv

EEG foundation models (EFMs) achieve strong cross-subject and cross-task generalization through large-scale pretraining and downstream fine-tuning. Through empirical analysis, we observe that (i) task-adapted EFMs provid…

Knowledge Distillation

PADD: Path-Aligned Decompression Distillation for Non-Router Teacher to Guide MoE Student Learning

2026-06-09 · Xinyue Peng, Yi Qian, Jiaojiao Lin, Wenjian Shao 외 arxiv

As large language models (LLMs) continue to scale, it becomes increasingly challenging to grow model capacity under fixed computation budgets. We propose Path-Aligned Decompression Distillation (PADD), a framework for di…

Knowledge DistillationMathematical Reasoning

AlphaRoute: Large Language Models as Semantic Optimizers for Multi-Objective Routing

2026-07-22 · Kabir Murjani, Mishri Bhavsar, Manish I. Patel, Jonti Talukdar arxiv

Very Large Scale Integration (VLSI) global routing is an NP-hard combinatorial optimization problem requiring signal net assignment across capacity-constrained 3D grids while minimizing congestion, wirelength, and via tr…