Corpus ID: 235485206

Unsupervised Resource Allocation with Graph Neural Networks

  title={Unsupervised Resource Allocation with Graph Neural Networks},
  author={M. Cranmer and Peter Melchior and Brian Nord},
We present an approach for maximizing a global utility function by learning how to allocate resources in an unsupervised way. We expect interactions between allocation targets to be important and therefore propose to learn the reward structure for near-optimal allocation policies with a GNN. By relaxing the resource constraint, we can employ gradient-based optimization in contrast to more standard evolutionary algorithms. Our algorithm is motivated by a problem in modern astronomy, where one… Expand

Figures and Tables from this paper

Graph Neural Network-based Resource Allocation Strategies for Multi-Object Spectroscopy
  • Tianshu Wang, P. Melchior
  • Computer Science, Physics
  • ArXiv
  • 2021
This work presents a bipartite Graph Neural Network architecture for trainable resource allocation strategies that enables fast adjustment and deployment of allocation strategies, statistical analyses of allocation patterns, and fully differentiable, science-driven solutions for resource allocation problems. Expand


Policy-GNN: Aggregation Optimization for Graph Neural Networks
Policy-GNN is proposed, a meta-policy framework that models the sampling procedure and message passing of GNNs into a combined learning process and significantly outperforms the state-of-the-art alternatives, showing the promise in aggregation optimization for GNN's. Expand
Resource Allocation via Graph Neural Networks in Free Space Optical Fronthaul Networks
The primal-dual learning algorithm is developed to train the graph neural network in a model-free manner, where the knowledge of system models is not required, and the GNN is shown to retain the permutation equivariance that matches with the permutations equivariances of resource allocation policy in networks. Expand
Resource allocation in the grid using reinforcement learning
The results of the experiments suggest that reinforcement learning can be used to improve the quality of resource allocation in large scale heterogenous system. Expand
NerveNet: Learning Structured Policy with Graph Neural Networks
NerveNet is proposed to explicitly model the structure of an agent, which naturally takes the form of a graph, and is demonstrated to be significantly more transferable and generalizable than policies learned by other models and are able to transfer even in a zero-shot setting. Expand
Mastering the game of Go with deep neural networks and tree search
Using this search algorithm, the program AlphaGo achieved a 99.8% winning rate against other Go programs, and defeated the human European Go champion by 5 games to 0.5, the first time that a computer program has defeated a human professional player in the full-sized game of Go. Expand
On the design of consequential ranking algorithms
This paper introduces a joint representation of rankings and user dynamics using Markov decision processes and shows that this representation greatly simplifies the construction of consequential ranking models that trade off the immediate utility and the long-term welfare. Expand
Deep Reinforcement Learning Based Resource Allocation for V2V Communications
From the simulation results, each agent can effectively learn to satisfy the stringent latency constraints on V2V links while minimizing the interference to vehicle-to-infrastructure communications. Expand
Differentiation of Blackbox Combinatorial Solvers
This work presents a method that implements an efficient backward pass through blackbox implementations of combinatorial solvers with linear objective functions, and incorporates the Gurobi MIP solver, Blossom V algorithm, and Dijkstra's algorithm into architectures that extract suitable features from raw inputs for the traveling salesman problem, the min-cost perfect matching problem and the shortest path problem. Expand
Learning Symbolic Physics with Graph Networks
An approach for imposing physically motivated inductive biases on graph networks to learn interpretable representations and improved zero-shot generalization is introduced and offers a valuable technique for interpreting and inferring explicit causal theories about the world from implicit knowledge captured by deep learning. Expand
A Framework for Telescope Schedulers: With Applications to the Large Synoptic Survey Telescope
A generic version of the Feature-Based scheduler, with minimal manual tailoring, is presented to demonstrate its potential and flexibility as a foundation for large ground-based telescope schedulers which can later be adjusted for other instruments. Expand