Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Robust Noise Attenuation via Adaptive Pooling of Transformer Outputs
Authors: Greyson Brothers
ICML 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our theoretical results are first validated by supervised experiments on a synthetic dataset designed to isolate the SNR problem, then generalized to standard relational reasoning, multi-agent reinforcement learning, and vision benchmarks with noisy observations, where transformers with adaptive pooling display superior robustness across tasks. |
| Researcher Affiliation | Academia | 1Johns Hopkins University Applied Physics Laboratory, Maryland, USA. Correspondence to: Greyson Brothers <EMAIL>. |
| Pseudocode | No | The paper describes methods, theorems, and proofs using mathematical notation and textual explanations, but it does not include any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | Code is publicly available at https://github.com/agbrothers/pooling. |
| Open Datasets | Yes | We use the standard Multi-Particle Environment (MPE) benchmark. ... Box World is a vision-based relational reasoning task introduced by Zambaldi et al. (2019). ... We conducted additional studies on image classification using the CIFAR 10 and 100 benchmark datasets. |
| Dataset Splits | Yes | For each of these pairs, we use 5-fold cross-validation with a holdout test set of 100k samples. ... All experiments trained vision transformers (Vi T) from scratch on the CIFAR dataset using 5-fold cross-validation. |
| Hardware Specification | No | The paper does not provide specific hardware details such as exact GPU/CPU models, processor types, or memory amounts used for running the experiments. |
| Software Dependencies | Yes | Our dataset was generated using Num Py version 2.0.2 with a seed of 42. |
| Experiment Setup | Yes | Additional hyperparameters are listed in the appendix C.2. ... Additional training hyperparameters can be found in C.4. ... Training hyperparameters and network architecture details are outlined in Table 16. |