Inductive Bias-driven Reinforcement Learning For Efficient Schedules in Heterogeneous Clusters

2020-01-01 · ICML 2020 1 · Subho Banerjee, Saurabh Jha, Zbigniew Kalbarczyk, Ravishankar Iyer

The problem of scheduling of workloads onto heterogeneous processors (e.g., CPUs, GPUs, FPGAs) is of fundamental importance in modern datacenters. Current system schedulers rely on application/system-specific heuristics that have to be built on a case-by-case basis. Recent work has demonstrated ML techniques to automate this heuristic search using black box approaches which require significant training data and time, which make them challenging to use in practice. This paper addresses the challenge in two ways: (i) a domain-driven Bayesian reinforcement learning (RL) model for scheduling, which inherently models the resource dependencies identified from the system architecture; and (ii) a sampling-based technique which allows the computation of gradients of a Bayesian model without performing full probabilistic inference. Together, these techniques reduce both the amount of training-data and -time required to produce scheduling policies that significantly outperform black box approaches by up to 2.2$\times$.

📄 PDF Abstract BibTeX

Code (0)

등록된 구현이 없습니다.

Tasks

Heuristic SearchInductive Biasreinforcement-learning Reinforcement Learning (RL)Scheduling

Inductive-bias-driven Reinforcement Learning For Efficient Schedules in Heterogeneous Clusters

2019-09-04 · Subho S. Banerjee, Saurabh Jha, Zbigniew T. Kalbarczyk, Ravishankar K. Iyer

Heuristic Search Inductive Bias reinforcement-learning Reinforcement Learning+2

Trajectory Entropy Reinforcement Learning for Predictable and Robust Control

2025-05-07 · Bang You, Chenxu Wang, Huaping Liu

Simplicity is a critical inductive bias for designing data-driven controllers, especially when robustness is important. Despite the impressive results of deep reinforcement learning in complex control tasks, it is prone …

Deep Reinforcement Learning Inductive Bias reinforcement-learning Reinforcement Learning

Inductive Bias for Emergent Communication in a Continuous Setting

2023-06-06 · John Isak Fjellvang Villanger, Troels Arnfred Bojesen

We study emergent communication in a multi-agent reinforcement learning setting, where the agents solve cooperative tasks and have access to a communication channel. The communication channel may consist of either discre…

Inductive Bias Multi-agent Reinforcement Learning reinforcement-learning Reinforcement Learning

In-Context Function Learning in Large Language Models

2026-02-12 · Elif Akata, Konstantinos Voudouris, Vincent Fortuin, Eric Schulz arxiv

Large language models (LLMs) can learn from a few demonstrations provided at inference time. We study this in-context learning phenomenon through the lens of Gaussian Processes (GPs). We build controlled experiments wher…

Reinforcement Learning Gaussian Processes

MLR-SNet: Transferable LR Schedules for Heterogeneous Tasks

2020-07-29 · Jun Shu, Yanwen Zhu, Qian Zhao, Zongben Xu 외

The learning rate (LR) is one of the most important hyper-parameters in stochastic gradient descent (SGD) algorithm for training deep neural networks (DNN). However, current hand-designed LR schedules need to manually pr…

text-classification Text Classification