End-to-end Stochastic Optimization with Energy-based Model

Authors: Lingkai Kong, Jiaming Cui, Yuchen Zhuang, Rui Feng, B. Aditya Prakash, Chao Zhang

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate SO-EBM in three applications: power scheduling, COVID-19 resource allocation, and non-convex adversarial security game, demonstrating the effectiveness and efficiency of SO-EBM.
Researcher Affiliation Academia College of Computing Georgia Institute of Technology {lkkong,jiamingcui1997,yczhuang,rfeng,badityap,chaozhang}@gatech.edu
Pseudocode Yes The detailed training procedure is given in Alg. 1 in the supplementary.
Open Source Code No The paper mentions providing supplementary material and an algorithm in the supplementary, but it does not contain an explicit statement like 'Our code is available at...' or 'We release our source code' nor does it provide a direct link to a code repository in the main text.
Open Datasets No The paper describes the characteristics of the data used for different tasks (e.g., '150-dimensional vector including the past day s electrical load and temperature' for power scheduling, 'SEIR+HD ODE model' for COVID-19), but it does not provide concrete access information (like a URL, DOI, or specific dataset name with citation) for a publicly available or open dataset.
Dataset Splits No The paper does not explicitly provide specific training/validation/test dataset splits (e.g., percentages, sample counts, or references to predefined splits) in the main text needed for reproducibility.
Hardware Specification No The acknowledgements section states 'funds/computing resources from Georgia Tech.' This is a general statement and does not provide specific details on the hardware used, such as exact GPU or CPU models, memory, or other detailed computer specifications for running the experiments.
Software Dependencies No The paper mentions using specific software tools such as 'QPTH solver [13]', 'cvxpylayers [1]', 'CVXPY [12, 2]', and 'Pyomo [5, 21]'. However, it does not specify version numbers for these software components, which are crucial for reproducible setup.
Experiment Setup Yes The paper provides concrete experimental setup details, such as specific hyperparameter values ('γs = 0.4, γe = 50 and c = 0.4' in Section 4.1), architectural choices ('2-hidden layer neural network' for forecasting in Section 4.1, 'two-layer gated recurrent unit (GRU)' in Section 4.2), and other settings ('sample from the forecasted distribution of β 100 times' in Section 4.2).