Janis Keuper, Paul Gavrikov
We lifted 9 functions out of this paper's own repositories and ran 6 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| paulgavrikov/biases_vs_generalization | canonical | 6 of 9 |
| Function | Status | Where it lives |
|---|---|---|
| bandpass_filter | Ran | paulgavrikov/biases_vs_generalization/code/eval_spectral.py pointer only (licence: CC-BY-SA-4.0) · get_code("1468a1f57586ad4d") |
| contrast_normalize_np | Ran | paulgavrikov/biases_vs_generalization/code/eval_criticalband.py pointer only (licence: CC-BY-SA-4.0) · get_code("e8501bbc645028fa") |
| deque_loader | Ran | paulgavrikov/biases_vs_generalization/code/eval_imagenet_adv.py pointer only (licence: CC-BY-SA-4.0) · get_code("71514d366caae3fe") |
| parse_aa_log | Ran | paulgavrikov/biases_vs_generalization/code/eval_imagenet_adv.py pointer only (licence: CC-BY-SA-4.0) · get_code("f96ad5482ab2b9ba") |
| r50_tf_to_torch | Ran | paulgavrikov/biases_vs_generalization/code/model_zoo.py pointer only (licence: CC-BY-SA-4.0) · get_code("b1945a1c081f47c5") |
| rmspower | Ran | paulgavrikov/biases_vs_generalization/code/eval_criticalband.py pointer only (licence: CC-BY-SA-4.0) · get_code("54537b5da9539618") |
| find_run_id_by_name | Not yet run | paulgavrikov/biases_vs_generalization/code/eval_imagenet_c.py pointer only (licence: CC-BY-SA-4.0) · get_code("2c006f49031dde59") |
| get_normalized_model | Not yet run | paulgavrikov/biases_vs_generalization/code/model_zoo.py pointer only (licence: CC-BY-SA-4.0) · get_code("461c3ff19b612b52") |
| load_state_dict_from_gdrive | Not yet run | paulgavrikov/biases_vs_generalization/code/model_zoo.py pointer only (licence: CC-BY-SA-4.0) · get_code("f4f66628816fbe4e") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
The robust generalization of models to rare, indistribution (ID) samples drawn from the long tail of the training distribution and to out-of-training-distribution (OOD) samples is one of the major challenges of current deep learning methods. For image classification, this manifests in the existence of adversarial attacks, the performance drops on distorted images, and a lack of generalization to concepts such as sketches. The current understanding of generalization in neural networks is very limited, but some biases that differentiate models from human vision have been identified and might be causing these limitations. Consequently, several attempts with varying success have been made to reduce these biases during training to improve generalization. We take a step back and sanitycheck these attempts. Fixing the architecture to the wellestablished ResNet-50, we perform a large-scale study on 48 ImageNet models obtained via different training methods to understand how and if these biases -including shape bias, spectral biases, and critical bands -interact with generalization. Our extensive study results reveal that contrary to previous findings, these biases are insufficient to accurately predict the generalization of a model holistically. We provide access to all checkpoints and evaluation code at https://github.com/paulgavrikov/biases_ vs_generalization/
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2404.01509")
get_code_for_paper("2404.01509")
have("2404.01509")
Connect an agent — have() is free.