Papers › Investigating and Simplifying Masking-based Saliency Methods for Model Interpretability

Investigating and Simplifying Masking-based Saliency Methods for Model Interpretability

19 Oct 2020arXiv:2010.09750archive 2025-07-28

Jason Phang, Jungkyu Park, Krzysztof J. Geras

Saliency maps that identify the most informative regions of an image for a classifier are valuable for model interpretability. A common approach to creating saliency maps involves generating input masks that mask out portions of an image to maximally deteriorate classification performance, or mask in an image to preserve classification performance. Many variants of this approach have been proposed in the literature, such as counterfactual generation and optimizing over a Gumbel-Softmax distribution. Using a general formulation of masking-based saliency methods, we conduct an extensive evaluation study of a number of recently proposed variants to understand which elements of these methods meaningfully improve performance. Surprisingly, we find that a well-tuned, relatively simple formulation of a masking-based saliency model outperforms many more complex approaches. We find that the most important ingredients for high quality saliency map generation are (1) using both masked-in and masked-out objectives and (2) training the classifier alongside the masking model. Strikingly, we show that a masking model can be trained with as few as 10 examples per class and still generate saliency maps with only a 0.7-point increase in localization error.

PaperPDFConference PDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 19 code samples harvested from 1 repository linked to this paper: 17 ran (of which 0 constructed an object rather than computing a result; 15 with no instrument failure: 0 honoured, 0 violated, 15 with no contract checked; 2 where Syntology's instrument failed) · 2 unverified. Where Syntology's instrument failed: 1 where our contract draft was wrong; 1 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 19 samples from 1 repository, 17 ran (15 with no instrument failure; 2 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written between 2026-09-25 and 2026-09-26 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

zphang/saliency_investigation officialmentioned in papermentioned on GitHubpytorchBSD-3-ClauseSyntology: 19 samples, 17 ran (2 instrument-failed) · 2 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

19 samples harvested: 17 ran (of which 0 constructed an object rather than computing a result; 15 with no instrument failure: 0 honoured, 0 violated, 15 with no contract checked; 2 where Syntology's instrument failed) · 2 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · our draft was wrong
1ran; the contract check could not run
15ran
2unverified

Licence: 1 of the 19 samples is pointer only, meaning Syntology does not serve that copy's text. For that sample, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from zphang/saliency_investigation. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

zphang/saliency_investigation official repository 19 samples, 17 ran (2 instrument-failed) · 2 unverified
binary_gumbel_softmax casme/archs.py ran BSD-3-Clause (permissive) · c75a899b50d166b6 · report fixture: signature logits:Tensor/2/float, tau:float/0/float, hard:bool/0/bool, epsilon:float/0/float · returned Tensor · level recorded 2026-09-25
default_infill_func casme/criterion.py ran fingerprinted BSD-3-Clause (permissive) · 8f29059810883f70 · report fixture: signature x:Tensor/4/float, mask:Tensor/4/float, generated_image:Tensor/4/float · returned Tensor · level recorded 2026-09-25
get_activation in 218 papers casme/ext/dfnet.py ran BSD-3-Clause (permissive) · ccefd8f533a37c96 · report fixture: signature name:str/0/str · returned ReLU · level recorded 2026-09-25
get_mask in 119 papers
census: 9 implementations from 10 papers, 9 distinct outputs across 7 buckets, one shared input per bucket
casme/model_basics.py ran BSD-3-Clause (permissive) · a2f9db8c0650f2df · report fixture: signature input_:Tensor/2/float, model:dict/0/?, use_p:NoneType/0/?, class_ids:NoneType/0/?, get_output:bool/0/bool, no_sigmoid:bool/0/bool · returned Tensor · level recorded 2026-09-25
get_masked_images casme/casme_utils.py ran BSD-3-Clause (permissive) · 544506b6b4e94b6e · report fixture: signature images:Tensor/4/float, binary_mask:Tensor/4/float, gray_scale:int/0/int, return_mask:bool/0/bool · returned tuple · level recorded 2026-09-25
get_masked_images_v2 casme/casme_utils.py ran fingerprinted BSD-3-Clause (permissive) · b257e814f0b02a0e · report fixture: signature batch_x:Tensor/2/float, mask:Tensor/2/float · returned tuple · level recorded 2026-09-25
get_masks_and_check_predictions casme/model_basics.py ran BSD-3-Clause (permissive) · 1eef12cb27f7130e · report fixture: signature input_:Tensor/4/float, target:Tensor/1/int, model:dict/0/?, erode_k:int/0/int, dilate_k:int/0/int, use_p:NoneType/0/?, no_sigmoid:bool/0/bool · returned tuple · level recorded 2026-09-25
get_norm in 48 papers casme/ext/dfnet.py ran BSD-3-Clause (permissive) · 22f472dba4017016 · report fixture: signature name:str/0/str, out_channels:int/0/int · returned BatchNorm2d · level recorded 2026-09-25
gram_matrix in 45 papers
census: 7 implementations from 9 papers, 4 distinct outputs on one shared input
casme/ext/pytorch_inpainting_with_partial_conv.py ran fingerprinted BSD-3-Clause (permissive) · 1715f300a6a58899 · report fixture: signature feat:Tensor/4/float · returned Tensor · level recorded 2026-09-25
ignore_superclass_signature casme/core.py ran BSD-3-Clause (permissive) · 9c25a48e312acd33 · report fixture: signature f:callable/0/? · returned function · level recorded 2026-09-25
mask_to_stats casme/stats.py ran fingerprinted BSD-3-Clause (permissive) · a29a81ff6ff8a73e · report fixture: signature mask:Tensor/4/float · returned dict · level recorded 2026-09-25
random_mask in 13 papers casme/ext/pconv_keras.py ran BSD-3-Clause (permissive) · ab33a7e0d32b0259 · report fixture: signature height:int/0/int, width:int/0/int, channels:int/0/int, min_multiplier:int/0/int, max_multiplier:int/0/int, number:int/0/int · returned ndarray · level recorded 2026-09-25
replace_prefix casme/archs.py ran fingerprinted BSD-3-Clause (permissive) · afc608bf21aa9f05 · report fixture: signature s:str/0/str, prefix:str/0/str · returned str · level recorded 2026-09-25
resize_like in 5 papers casme/ext/dfnet.py ran BSD-3-Clause (permissive) · 72d2ca5c8b19b945 · report fixture: signature x:Tensor/4/float, target:Tensor/4/float, mode:str/0/str · returned Tensor · level recorded 2026-09-25
resnet50shared casme/archs.py ran BSD-3-Clause (permissive) · f067739c23018f46 · report fixture: signature pretrained:bool/0/bool, path:NoneType/0/? · returned ResNet · level recorded 2026-09-25
total_variation_loss in 18 papers casme/ext/pytorch_inpainting_with_partial_conv.py ran; the contract check could not run (RAISES) fingerprinted BSD-3-Clause (permissive) · dedfa1b69056bdc7 · report fixture: signature image:Tensor/4/float · returned Tensor · level recorded 2026-09-26
weights_init in 44 papers casme/ext/pytorch_inpainting_with_partial_conv.py ran · our draft was wrong BSD-3-Clause (permissive) · b0a0259ffdb9c7d8 · report fixture: signature init_type:str/0/str · returned function · level recorded 2026-09-26
accuracy in 872 papers
census: 8 implementations from 7 papers, 7 distinct outputs across 6 buckets, one shared input per bucket
casme/train_utils.py unverified BSD-3-Clause (permissive) here; the stored copy is pointer only · 688c38ac8517471a · report
apply_uniform_random_value_mask_func casme/criterion.py unverified BSD-3-Clause (permissive) · ca4396c9e5d986b8 · report

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections