No One Representation to Rule Them All: Overlapping Features of Training Methods

• URL: http://arxiv.org/abs/2110.12899v2
• Date: Tue, 26 Oct 2021 17:29:51 GMT
• Title: No One Representation to Rule Them All: Overlapping Features of Training Methods
• Authors: Raphael Gontijo-Lopes, Yann Dauphin, Ekin D. Cubuk
• Abstract summary: High-performing models tend to make similar predictions regardless of training methodology. Recent work has made very different training techniques, such as large-scale contrastive learning, yield competitively-high accuracy. We show these models specialize in generalization of the data, leading to higher ensemble performance.
• Score: 12.58238785151714
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Despite being able to capture a range of features of the data, high accuracy models trained with supervision tend to make similar predictions. This seemingly implies that high-performing models share similar biases regardless of training methodology, which would limit ensembling benefits and render low-accuracy models as having little practical use. Against this backdrop, recent work has made very different training techniques, such as large-scale contrastive learning, yield competitively-high accuracy on generalization and robustness benchmarks. This motivates us to revisit the assumption that models necessarily learn similar functions. We conduct a large-scale empirical study of models across hyper-parameters, architectures, frameworks, and datasets. We find that model pairs that diverge more in training methodology display categorically different generalization behavior, producing increasingly uncorrelated errors. We show these models specialize in subdomains of the data, leading to higher ensemble performance: with just 2 models (each with ImageNet accuracy ~76.5%), we can create ensembles with 83.4% (+7% boost). Surprisingly, we find that even significantly low-accuracy models can be used to improve high-accuracy models. Finally, we show diverging training methodology yield representations that capture overlapping (but not supersetting) feature sets which, when combined, lead to increased downstream performance.

Related papers

• A Hitchhiker's Guide to Scaling Law Estimation [56.06982415792523]
  Scaling laws predict the loss of a target machine learning model by extrapolating from easier-to-train models with fewer parameters or smaller training sets. We estimate more than 1000 scaling laws, then derive a set of best practices for estimating scaling laws in new model families.
  arXiv  Detail & Related papers  (2024-10-15T17:59:10Z)
• Fantastic Gains and Where to Find Them: On the Existence and Prospect of General Knowledge Transfer between Any Pretrained Model [74.62272538148245]
  We show that for arbitrary pairings of pretrained models, one model extracts significant data context unavailable in the other. We investigate if it is possible to transfer such "complementary" knowledge from one model to another without performance degradation.
  arXiv  Detail & Related papers  (2023-10-26T17:59:46Z)
• A study on the impact of pre-trained model on Just-In-Time defect prediction [10.205110163570502]
  We build six models: RoBERTaJIT, CodeBERTJIT, BARTJIT, PLBARTJIT, GPT2JIT, and CodeGPTJIT, each with a distinct pre-trained model as its backbone. We investigate the performance of the models when using Commit code and Commit message as inputs, as well as the relationship between training efficiency and model distribution.
  arXiv  Detail & Related papers  (2023-09-05T15:34:22Z)
• Feature Weaken: Vicinal Data Augmentation for Classification [1.7013938542585925]
  We use Feature Weaken to construct the vicinal data distribution with the same cosine similarity for model training. This work can not only improve the classification performance and generalization of the model, but also stabilize the model training and accelerate the model convergence.
  arXiv  Detail & Related papers  (2022-11-20T11:00:23Z)
• Investigating Ensemble Methods for Model Robustness Improvement of Text Classifiers [66.36045164286854]
  We analyze a set of existing bias features and demonstrate there is no single model that works best for all the cases. By choosing an appropriate bias model, we can obtain a better robustness result than baselines with a more sophisticated model design.
  arXiv  Detail & Related papers  (2022-10-28T17:52:10Z)
• Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time [69.7693300927423]
  We show that averaging the weights of multiple models fine-tuned with different hyper parameter configurations improves accuracy and robustness. We show that the model soup approach extends to multiple image classification and natural language processing tasks.
  arXiv  Detail & Related papers  (2022-03-10T17:03:49Z)
• Exploring Strategies for Generalizable Commonsense Reasoning with Pre-trained Models [62.28551903638434]
  We measure the impact of three different adaptation methods on the generalization and accuracy of models. Experiments with two models show that fine-tuning performs best, by learning both the content and the structure of the task, but suffers from overfitting and limited generalization to novel answers. We observe that alternative adaptation methods like prefix-tuning have comparable accuracy, but generalize better to unseen answers and are more robust to adversarial splits.
  arXiv  Detail & Related papers  (2021-09-07T03:13:06Z)
• The Evolution of Out-of-Distribution Robustness Throughout Fine-Tuning [25.85044477227461]
  Models that are more accurate on the out-of-distribution data relative to this baseline exhibit "effective robustness" We find that models pre-trained on larger datasets exhibit effective robustness during training that vanishes at convergence. We discuss several strategies for scaling effective robustness to the high-accuracy regime to improve the out-of-distribution accuracy of state-of-the-art models.
  arXiv  Detail & Related papers  (2021-06-30T06:21:42Z)
• When Ensembling Smaller Models is More Efficient than Single Large Models [52.38997176317532]
  We show that ensembles can outperform single models with both higher accuracy and requiring fewer total FLOPs to compute. This presents an interesting observation that output diversity in ensembling can often be more efficient than training larger models.
  arXiv  Detail & Related papers  (2020-05-01T18:56:18Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.