Ensuring DNN Solution Feasibility for Optimization Problems with Linear Constraints
Authors: Tianyu Zhao, Xiang Pan, Minghua Chen, Steven Low
ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Simulation results over IEEE test cases show that it outperforms existing strong DNN baselines in ensuring 100% feasibility and attaining consistent optimality loss (<0.19%) and speedup (up to 228) in both light-load and heavy-load regimes, as compared to a state-of-the-art solver. |
| Researcher Affiliation | Collaboration | Tianyu Zhao1,2, Xiang Pan2, Minghua Chen3,*, Steven H. Low4 1Lenovo Machine Intelligence Center, 2The Chinese University of Hong Kong, 3City University of Hong Kong, 4California Institute of Technology |
| Pseudocode | Yes | Algorithm 1: Determining Sufficient DNN Size |
| Open Source Code | No | The paper does not contain an explicit statement about releasing its source code or a link to a code repository. |
| Open Datasets | Yes | We evaluate its performance over IEEE 30-/118-/300bus test cases (tpc, 2018) on the input load region of [100%, 130%] of the default load covering both the light-load ([100%, 115%]) and heavy-load ([115%, 130%]) regimes, respectively. (tpc, 2018) refers to Power Systems Test Case Archive, 2018. http://labs.ece.uw.edu/pstca/. |
| Dataset Splits | No | The paper mentions 'training set' and 'test setting' but does not provide specific details on training, validation, and test splits (e.g., percentages or exact counts) needed for reproduction. |
| Hardware Specification | Yes | We conduct simulations in Cent OS 7.6 with a quad-core (i7-3770@3.40G Hz) CPU and 16GB RAM. |
| Software Dependencies | No | The paper mentions software like 'Pypower' and 'Gurobi' but does not provide specific version numbers for these or any other key software dependencies. |
| Experiment Setup | Yes | The paper describes the loss function (Equation 11) with weighting factors w1 and w2. A footnote in Table 1 states: 'The training epochs for the other DNN schemes are 200'. |