- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper proposes EvoMIDFL, an agentic framework that evolves surrogate backward-gradient policies for mixed-integer decision-focused learning. The method combines several gradient estimators, validates candidate policies from the same training checkpoint, and maintains a three-layer memory for transferring successful configurations across scales and problem families. Experiments on bipartite matching and knapsack show that EvoMIDFL is often competitive with fixed backward methods, with small improvements in several settings and evidence of cross-task transfer.
Critical strengths
- The paper gives a coherent framework for selecting, configuring, and composing heterogeneous surrogate gradients.
- The memory-transfer ablations are useful.
Critical weaknesses
- The empirical gains are generally small and not accompanied by standard deviations, confidence intervals, or paired significance tests.
- The evolution process repeatedly selects policies using the same validation set over many rounds and hundreds of proposals, creating a substantial risk of validation overfitting.
- The contribution of the LLM itself is not isolated from a similarly budgeted non-LLM search or hyperparameter optimization baseline, and the computational/token cost is not reported.
Questions for authors
- Can you report per-seed results and statistical tests for the main comparisons, especially matching-225, matching-900, and knapsack-2000?
- How much wall-clock time, solver calls, and LLM/token usage does EvoMIDFL require compared with fixed baselines and an equally budgeted random or Bayesian policy search?
- How do the results change with an independent meta-validation set, or with a reversed transfer direction (knapsack to matching) and an unseen MILP family?
12 comments
Detailed Response to Reviewer LiAv (Weakness)
Weakness: LLM-driven benefits.
We thank the reviewer for pointing out the weakness. To thoroughly evaluate the contribution of LLM-driven evolution, we first conducted three new automated search experiments (AutoML methods): Random Search, Evolutionary Search, and Bayesian Optimization, and then clarified the benefits of LLM-driven methods.
1. New experiments of AutoML:
- Settings: In AutoML experiments, we replaced the Actor with each AutoML method in turn. Results demonstrate that under an identical budget of validation rounds, EvoMIDFL–Full achieves lower mean test regret than all three baselines across every evaluated setting (in Table 3: Comparison with conventional search baselines on Matching and Knapsack in the main paper).
- Empirical gains. Empirically, our quantitative results confirm that LLM-driven, diagnostic-guided policies yield superior decision quality compared to non-LLM baselines under an identical evolutionary budget. As shown in Table 3, given the same number of evolution rounds, EvoMIDFL–Full achieves the lowest mean test regret across all four Matching and Knapsack benchmarks. Relative to the strongest AutoML baseline on each task, the most substantial regret reductions occur on the Matching benchmarks:
7.4%on Matching-900 k-1(from47.862to44.304) and9.2%on Matching-1600. On Knapsac000 and -2000, the reductions in mean test regret per item are more modest, yet consistent at1.1%and1.5%, respectively.
2. LLM-driven benefits:
Based on the conventional AutoML results and previous comparison with fixed baselines, EvoMIDFL overcomes both (i) the blind black-box parameter tuning of conventional AutoML and (ii) the rigid, hand-crafted heuristics of fixed baselines that fail to adapt across problems or scales. The two key differences between EvoMIDFL and conventional AutoML search or fixed baselines are detailed below:
(1) Diagnostic-Guided Reasoning (EvoMIDFL v.s. AutoML)
- How EvoMIDFL works. The LLM Actor agent explicitly generates proposals based on training diagnostics, the parent policy, and recent trial outcomes. For example, when validation regret plateaus alongside fluctuating gradient norms, the Actor can diagnose optimization instability and prioritize a coordinated candidate, such as switching the gradient estimator while simultaneously tightening the clipping threshold to directly resolve the underlying failure mode.
- Conventional search. In contrast, conventional AutoML methods treat policy search as a black-box optimization problem relying solely on scalar validation scores. Specifically, Random Search samples uniformly, Evolutionary Search perturbs candidates near historically strong policies, and Bayesian Optimization fits a surrogate model exclusively on past validation scores without inspecting intermediate training dynamics. Crucially, conventional AutoML methods typically optimize scalar validation metrics via black-box sampling, lacking visibility into ongoing training dynamics to diagnose failure modes or perform targeted, structural policy refinements.
- Benefits: The targeted-validation diagnosis in EvoMIDFL overcomes the blind black-box parameter tuning of conventional AutoML, which is supported by the empirical experiments of EvoMIDFL with AutoML under identical evolution rounds.
(2) Paradigm Shift Toward Dynamic, Training-Aware Policy Adaptation: (EvoMIDFL v.s. fixed baselines)
-
Conventional DFL: Conventional decision-focused learning (DFL) pipelines typically treat the backward gradient estimator as a static design choice fixed prior to training, requiring practitioners to manually select, tune, and validate candidate surrogate formulations through laborious trial and error.
-
New paradigm: EvoMIDFL shifts this paradigm by formulating gradient-policy selection as a dynamic, sequential decision process that continuously adapts alongside evolving training dynamics. By tightly coupling structured policy proposals with validation-gated promotion, the framework autonomously navigates combinatorial spaces of estimators, hyperparameters, and stabilization mechanisms. This closed-loop automation eliminates the need for exhaustive manual candidate enumeration and domain-specific hand-crafting, substantially reducing human engineering effort while adapting seamlessly to varying optimization stages and problem structures.
-
Balancing accuracy and stability: EvoMIDFL ensures both high accuracy and strong stability in final decision quality. Across the majority of evaluated settings, EvoMIDFL outperforms individual single-policy baselines in terms of test regret in most occasions; even in the few scenarios where performance gains are more modest, the margins remain small and competitive, demonstrating reliable cross-scenario robustness without significant performance degradation.
Detailed Response to Reviewer LiAv (Weakness cont.)
Table 3: Comparison with conventional search baselines on Matching and Knapsack (in the main paper). RS, ES, and BO denote random search, evolutionary search, and Bayesian optimization. Knapsack regret is per item.
| Method | Matching-900 regret ↓ | Matching-900 RMSE ↓ | Matching-1600 regret ↓ | Matching-1600 RMSE ↓ | Knapsack-1000 regret/item ↓ | Knapsack-1000 RMSE ↓ | Knapsack-2000 regret/item ↓ | Knapsack-2000 RMSE ↓ |
|---|---|---|---|---|---|---|---|---|
| RS | 51.254 ± 0.704 | 8.736 ± 1.604 | 72.654 ± 6.041 | 9.620 ± 0.947 | 0.289 ± 0.022 | 2.260 ± 0.359 | 0.304 ± 0.064 | 2.439 ± 0.647 |
| ES | 49.559 ± 4.863 | 8.441 ± 2.154 | 70.473 ± 5.480 | 9.742 ± 0.831 | 0.266 ± 0.011 | 1.885 ± 0.005 | 0.265 ± 0.009 | 2.174 ± 0.132 |
| BO | 47.862 ± 1.533 | 9.372 ± 1.106 | 72.429 ± 2.040 | 7.126 ± 1.251 | 0.270 ± 0.023 | 1.979 ± 0.127 | 0.268 ± 0.004 | 3.604 ± 0.399 |
| EvoMIDFL–Full (Ours) | 44.304 ± 0.364 | 7.204 ± 0.431 | 64.011 ± 0.333 | 6.939 ± 0.978 | 0.263 ± 0.035 | 1.873 ± 0.218 | 0.261 ± 0.024 | 2.025 ± 0.237 |
General Response to Reviewer LiAv
We thank the reviewer for the thoughtful evaluation and positive remarks regarding our surrogate-gradient adaptation framework and memory ablations.
Guided by your suggestions, we have expanded our experimental analysis in the revised manuscript as follows:
We are grateful for these recommendations, which have allowed us to demonstrate the rigor and practical scope of EvoMIDFL more comprehensively. Detailed discussions and experimental evidence for each point are provided below.