SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.01875 · NeurIPS · 2023

Towards Stable Backdoor Purification through Feature Shift Tuning

Minhao Cheng, Zeyu Qin, Li Shen, Rui Min

arXiv · PDF · Open in the Atlas

Code that ran

10 of 13 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 4 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 5 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2310.01875.json · badge: /api/ran/2310.01875.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
AISafety-HKUST/stable_backdoor_purification canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
copy not recorded community
not in the paper's declared repositories
2 ran · 0 no recorded run
aisafety-hkust/stable_backdoor_purification alias (classifier label)
classifier basis: classifier_2025
1 ran · 1 no recorded run
kr-anonymous/ngf-animus reimplementation (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
FunctionStatusWhere it lives
add_common_attack_args Ran
ran; our contract draft was wrong, not the code
AISafety-HKUST/stable_backdoor_purification/attack/badnet.py
pointer only (licence: NONE); the stored copy get_code serves, from retsuh-bqw/fmp, is pointer only (licence: NOASSERTION) · get_code("efd3b8b851a0f5e7")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, AISafety-HKUST/stable_backdoor_purification when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

add_common_attack_args Ran
ran; our contract draft was wrong, not the code
AISafety-HKUST/stable_backdoor_purification/attack/badnet_bypass.py
pointer only (licence: NOASSERTION) · get_code("5cfe5054c2c74fce")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, AISafety-HKUST/stable_backdoor_purification (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

generalize_to_lower_pratio Ran
ran, honoured the contract we drafted
AISafety-HKUST/stable_backdoor_purification/attack/wanet.py
pointer only (licence: NONE); the stored copy get_code serves, from sclbd/backdoorbench, is pointer only (licence: NOASSERTION) · get_code("1e9b4bedd9340fe0")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, AISafety-HKUST/stable_backdoor_purification when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

generate_single_target_attack_train_pidx Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
AISafety-HKUST/stable_backdoor_purification/utils/backdoor_generate_pindex.py
pointer only (licence: NOASSERTION) · get_code("ee9472dc60824bd5")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, AISafety-HKUST/stable_backdoor_purification (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

generate_single_target_attack_train_poison_index Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
AISafety-HKUST/stable_backdoor_purification/utils/backdoor_generate_poison_index.py
pointer only (licence: NOASSERTION) · get_code("ad7ac1c21671d465")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, AISafety-HKUST/stable_backdoor_purification (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

is_valid_file Ran
ran, violated the contract we drafted
AISafety-HKUST/stable_backdoor_purification/for_imagenet/generate_poison_val_badnet.py
pointer only (licence: NONE); the stored copy get_code serves, from shawkui/Proactive_Defensive_Backdoor, is pointer only (licence: NOASSERTION) · get_code("ea3c60ddf40d46bc")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, AISafety-HKUST/stable_backdoor_purification when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

test Ran
ran, honoured the contract we drafted
this paper's copy was not recorded; identical code first harvested from csdongxian/ANP_backdoor
pointer only; the stored copy get_code serves, from csdongxian/ANP_backdoor, is pointer only (licence: NONE) · get_code("f634daea78f08e78")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

test Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
kr-anonymous/ngf-animus/Purification_NGF.py
pointer only (licence: NONE) · get_code("0044d4606bb15d8c")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kr-anonymous/ngf-animus when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

test Ran
ran, honoured the contract we drafted
aisafety-hkust/stable_backdoor_purification/fine_tune/BTIDBF/btidbfp.py
pointer only (licence: NOASSERTION) · get_code("5840d4aeba85a839")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, aisafety-hkust/stable_backdoor_purification (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

train Ran
ran, honoured the contract we drafted
this paper's copy was not recorded; identical code first harvested from csdongxian/ANP_backdoor
pointer only; the stored copy get_code serves, from csdongxian/ANP_backdoor, is pointer only (licence: NONE) · get_code("1042d89ceda24cdb")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

generate_pidx_from_label_transform Not yet run
unverified (harvested, no recorded run)
AISafety-HKUST/stable_backdoor_purification/utils/backdoor_generate_pindex.py
pointer only (licence: NOASSERTION) · get_code("83953e32e89f7c00")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, AISafety-HKUST/stable_backdoor_purification (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

generate_poison_index_from_label_transform Not yet run
unverified (harvested, no recorded run)
AISafety-HKUST/stable_backdoor_purification/utils/backdoor_generate_poison_index.py
pointer only (licence: NOASSERTION) · get_code("3c55a855cc155090")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, AISafety-HKUST/stable_backdoor_purification (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

purify Not yet run
unverified (harvested, no recorded run)
aisafety-hkust/stable_backdoor_purification/fine_tune/BTIDBF/btidbfp.py
pointer only (licence: NOASSERTION) · get_code("ca2a104dfcdcd492")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, aisafety-hkust/stable_backdoor_purification (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 5 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 0 the same month, 3 later) and 8 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
add_common_attack_args 8 other papers
6 dated earlier, 2 later
1708.06733 (2017-08)
BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain
1905.13409 (2019-05)
Bypassing Backdoor Detection Algorithms in Deep Learning
2206.12654 (2022-06)
BackdoorBench: A Comprehensive Benchmark of Backdoor Learning
and 5 more not named here
generalize_to_lower_pratio 4 other papers
2 dated earlier, 2 later
2307.10562 (2023-07)
Shared Adversarial Unlearning: Backdoor Mitigation by Unlearning Shared Adversarial Examples
2307.11565 (2023-07)
Adversarial Feature Map Pruning for Backdoor
2405.16112 (2024-05)
Mitigating Backdoor Attack by Injecting Proactive Defensive Backdoor
and 1 more not named here
is_valid_file 3 other papers
2 dated earlier, 1 later
2307.10562 (2023-07)
Shared Adversarial Unlearning: Backdoor Mitigation by Unlearning Shared Adversarial Examples
2307.11565 (2023-07)
Adversarial Feature Map Pruning for Backdoor
2405.16112 (2024-05)
Mitigating Backdoor Attack by Injecting Proactive Defensive Backdoor
test 2 other papers
2 dated earlier
2110.14430 (2021-10)
Adversarial Neuron Pruning Purifies Backdoored Deep Models
2305.14876 (2023-05)
Reconstructive Neuron Pruning for Backdoor Defense
train 1 other paper
1 dated earlier
2110.14430 (2021-10)
Adversarial Neuron Pruning Purifies Backdoored Deep Models

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

It has been widely observed that deep neural networks (DNN) are vulnerable to backdoor attacks where attackers could manipulate the model behavior maliciously by tampering with a small set of training samples. Although a line of defense methods is proposed to mitigate this threat, they either require complicated modifications to the training process or heavily rely on the specific model architecture, which makes them hard to deploy into real-world applications. Therefore, in this paper, we instead start with fine-tuning, one of the most common and easy-to-deploy backdoor defenses, through comprehensive evaluations against diverse attack scenarios. Observations made through initial experiments show that in contrast to the promising defensive results on high poisoning rates, vanilla tuning methods completely fail at low poisoning rate scenarios. Our analysis shows that with the low poisoning rate, the entanglement between backdoor and clean features undermines the effect of tuning-based defenses. Therefore, it is necessary to disentangle the backdoor and clean features in order to improve backdoor purification. To address this, we introduce Feature Shift Tuning (FST), a method for tuning-based backdoor purification. Specifically, FST encourages feature shifts by actively deviating the classifier weights from the originally compromised weights. Extensive experiments demonstrate that our FST provides consistently stable performance under different attack settings. Without complex parameter adjustments, FST also achieves much lower tuning costs, only 10 epochs. Our codes are available at https: //github.com/AISafety-HKUST/stable_backdoor_purification.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2310.01875")
get_code_for_paper("2310.01875")
have("2310.01875")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2310.01875.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2310.01875.svg)](https://syntology.ai/paper/2310.01875)

Connect an agent — have() is free.