Papers › Gradient Centralization: A New Optimization Technique for Deep Neural Networks

Gradient Centralization: A New Optimization Technique for Deep Neural Networks

3 Apr 2020ECCV 2020 8arXiv:2004.01461archive 2025-07-28

Hongwei Yong, Jianqiang Huang, Xian-Sheng Hua, Lei Zhang

Optimization techniques are of great importance to effectively and efficiently train a deep neural network (DNN). It has been shown that using the first and second order statistics (e.g., mean and variance) to perform Z-score standardization on network activations or weight vectors, such as batch normalization (BN) and weight standardization (WS), can improve the training performance. Different from these existing methods that mostly operate on activations or weights, we present a new optimization technique, namely gradient centralization (GC), which operates directly on gradients by centralizing the gradient vectors to have zero mean. GC can be viewed as a projected gradient descent method with a constrained loss function. We show that GC can regularize both the weight space and output feature space so that it can boost the generalization performance of DNNs. Moreover, GC improves the Lipschitzness of the loss function and its gradient so that the training process becomes more efficient and stable. GC is very simple to implement and can be easily embedded into existing gradient based DNN optimizers with only one line of code. It can also be directly used to fine-tune the pre-trained DNNs. Our experiments on various applications, including general image classification, fine-grained image classification, detection and segmentation, demonstrate that GC can consistently improve the performance of DNN learning. The code of GC can be found at https://github.com/Yonghongwei/Gradient-Centralization.

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

The archive's official repository (Yonghongwei/Gradient-Centralization) was not harvested; the 29 samples below come from 4 community repositories.

Syntology 29 code samples harvested from 4 repositories linked to this paper: 18 ran (of which 0 constructed an object rather than computing a result; 17 with no instrument failure: 1 honoured, 0 violated, 16 with no contract checked; 1 where Syntology's instrument failed) · 11 unverified. Where Syntology's instrument failed: 1 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: community (archive-listed): 29 samples from 4 repositories, 18 ran (17 with no instrument failure; 1 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.

No sample is linked to this paper from 2 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-22 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

Yonghongwei/Gradient-Centralization officialmentioned in papermentioned on GitHubpytorchSyntology: not harvested report
HamadYA/GhostFaceNets mentioned on GitHubtfMITSyntology: 22 samples, 16 ran (0 instrument-failed) · 6 unverified report
IssamLaradji/sps mentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Locally Adaptive Federated Learning (arXiv:2307.06306; that paper's own run record: 14 ran (of which 0 constructed an object rather than computing a result; 14 with no instrument failure: 0 honoured, 0 violated, 14 with no contract checked; 0 where Syntology's instrument failed) · 2 unverified); nothing checks that they implement this paper's method. report
Rishit-dagli/Gradient-Centralization-TensorFlow mentioned on GitHubtfApache-2.0Syntology: 5 samples, 0 ran · 5 unverified report
lessw2020/Ranger-Deep-Learning-Optimizer mentioned on GitHubpytorchApache-2.0Syntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified report
mhassann22/GCSAM mentioned on GitHubpytorchSyntology: 1 sample, 1 ran (1 instrument-failed) · 0 unverified report
mnikitin/Gradient-Centralization mentioned on GitHubmxnetSyntology: not harvested report
keras-team/keras-io tfSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to The Forward-Forward Algorithm: Some Preliminary Investigations (arXiv:2212.13345; that paper's own run record: 18 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 0 honoured, 2 violated, 9 with no contract checked; 7 where Syntology's instrument failed) · 2 unverified); nothing checks that they implement this paper's method. 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.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

29 samples harvested: 18 ran (of which 0 constructed an object rather than computing a result; 17 with no instrument failure: 1 honoured, 0 violated, 16 with no contract checked; 1 where Syntology's instrument failed) · 11 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · honoured contract
1ran; the contract check could not run
16ran
11unverified

Licence: 3 of the 29 samples are pointer only, meaning Syntology does not serve that copy's text. For 2 of them, 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 4 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “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.

HamadYA/GhostFaceNets community (archive-listed) 22 samples, 16 ran (0 instrument-failed) · 6 unverified
activation in 30 papers backbones/ghost_model.py ran MIT (permissive) · 619357c6cd6c6727 · report fixture: signature inputs:KerasTensor/0/float · returned KerasTensor · level recorded 2026-09-24
add_l2_regularizer_2_model GhostFaceNets.py ran MIT (permissive) · 719582a083732fb7 · report fixture: signature model:callable/0/?, weight_decay:float/0/float, custom_objects:dict/0/?, apply_to_batch_normal:bool/0/bool, apply_to_bias:bool/0/bool · returned Functional · level recorded 2026-09-24
buildin_models GhostFaceNets.py ran MIT (permissive) · e8ddfad281bed049 · report fixture: signature stem_model:callable/0/?, dropout:float/0/float, emb_shape:int/0/int, input_shape:tuple/1/int, output_layer:str/0/str, bn_momentum:float/0/float, bn_epsilon:float/0/float, add_pointwise_conv:bool/0/bool, pointwise_conv_act:str/0/str, use_bias:bool/0/bool, scale:bool/0/bool, weights:str/0/str · returned Functional · level recorded 2026-09-24
buildin_models GhostFaceNets_with_Bias.py ran MIT (permissive) · 6dbeec44cfb81d9f · report fixture: signature stem_model:callable/0/?, dropout:float/0/float, emb_shape:int/0/int, input_shape:tuple/1/int, output_layer:str/0/str, bn_momentum:float/0/float, bn_epsilon:float/0/float, add_pointwise_conv:bool/0/bool, pointwise_conv_act:str/0/str, use_bias:bool/0/bool, scale:bool/0/bool, weights:str/0/str · returned Functional · level recorded 2026-09-24
calculate_roc in 14 papers evals.py ran MIT (permissive) · c3af5327e1b4e163 · report fixture: signature thresholds:ndarray/1/float, embeddings1:ndarray/2/float, embeddings2:ndarray/2/float, actual_issame:ndarray/1/bool, nrof_folds:int/0/int, pca:int/0/int · returned tuple · level recorded 2026-09-25
distiller_loss_cosine losses.py ran fingerprinted MIT (permissive) · 4799f292d886421d · report fixture: signature true_emb:EagerTensor/2/float, pred_emb:EagerTensor/2/float · returned EagerTensor · level recorded 2026-09-24
distiller_loss_euclidean losses.py ran fingerprinted MIT (permissive) · 82eb5716083ae566 · report fixture: signature true_emb:EagerTensor/2/float, pred_emb:EagerTensor/2/float · returned EagerTensor · level recorded 2026-09-24
face_align_landmark IJB_evals.py ran MIT (permissive) · 95b7e550f0df874b · report fixture: signature img:ndarray/3/int, landmark:ndarray/2/float, image_size:tuple/1/int, method:str/0/str · returned ndarray · level recorded 2026-09-24
ghost_module backbones/ghost_model.py ran MIT (permissive) · 2dcab8fa77a49fd8 · report fixture: signature inputs:KerasTensor/0/float, out:int/0/int, convkernel:int/0/int, dwkernel:int/0/int, add_activation:bool/0/bool · returned KerasTensor · level recorded 2026-09-24
half_split_weighted_cosine_similarity_11 evals.py ran fingerprinted MIT (permissive) · 75f14459448a4cef · report fixture: signature aa:EagerTensor/2/float, bb:EagerTensor/2/float · returned EagerTensor · level recorded 2026-09-24
plot_tpr_far eval_folder.py ran MIT (permissive) · aa04aa5fb992d2bc · report fixture: signature score:ndarray/1/float, label:ndarray/1/int, new_figure:bool/0/bool, label_prefix:str/0/str · returned Figure · level recorded 2026-09-24
read_IJB_meta_columns_to_int IJB_evals.py ran MIT (permissive) · c5ac51f2fe4a6729 · report fixture: signature file_path:str/0/str, columns:list/1/int, sep:str/0/str, skiprows:int/0/int, header:NoneType/0/? · returned generator · level recorded 2026-09-24
replace_ReLU_with_PReLU GhostFaceNets.py ran MIT (permissive) · d1106a8957c8d037 · report fixture: signature model:callable/0/?, target_activation:str/0/str · returned Functional · level recorded 2026-09-24
se_module backbones/ghost_model.py ran MIT (permissive) · ed575936f30efada · report fixture: signature inputs:KerasTensor/0/float, se_ratio:float/0/float · returned KerasTensor · level recorded 2026-09-24
teacher_model_interf_wrapper data_distiller.py ran MIT (permissive) · 36a0054d36a367be · report fixture: signature model_file:str/0/str · returned function · level recorded 2026-09-24
to_4d in 69 papers augment.py ran · honoured contract fingerprinted MIT (permissive) here; the stored copy is pointer only · bd1c72a9d8556505 · report fixture: signature image:EagerTensor/2/float · returned EagerTensor · level recorded 2026-09-22
from_4d augment.py unverified MIT (permissive) here; the stored copy is pointer only · b10fb8ac53c53a97 · report
half_split_weighted_cosine_similarity evals.py unverified MIT (permissive) · b76d78b83d84097c · report
keras_model_interf IJB_evals.py unverified MIT (permissive) · 0d40d2078622e4a1 · report
random_cutout_or_cutout_mask data.py unverified MIT (permissive) · b0dc513e1e285780 · report
replace_add_with_stochastic_depth models.py unverified MIT (permissive) · 5de5a42ab87d80ec · report
tf_imread data.py unverified MIT (permissive) · 95d626160eefd668 · report
Rishit-dagli/Gradient-Centralization-TensorFlow community (archive-listed) 5 samples, 0 ran · 5 unverified
adadelta gctf/optimizers.py unverified Apache-2.0 (permissive) · 8f86e7549e63b17d · report
adagrad gctf/optimizers.py unverified Apache-2.0 (permissive) · a02310673d3edff8 · report
centralized_gradients_for_optimizer gctf/centralized_gradients.py unverified Apache-2.0 (permissive) · fd031cd7977c1eee · report
get_centralized_gradients gctf/centralized_gradients.py unverified Apache-2.0 (permissive) · 52780979cc1c6e50 · report
update_optimizer gctf/optimizers.py unverified Apache-2.0 (permissive) · df2465eb334d9036 · report
lessw2020/Ranger-Deep-Learning-Optimizer community (archive-listed) 1 sample, 1 ran (0 instrument-failed) · 0 unverified
centralized_gradient ranger/ranger2020.py ran fingerprinted Apache-2.0 (permissive) · 62a2acf5cdbc743f · report fixture: signature x:Tensor/2/float, use_gc:bool/0/bool, gc_conv_only:bool/0/bool · returned Tensor · level recorded 2026-09-25
mhassann22/GCSAM community (archive-listed) 1 sample, 1 ran (1 instrument-failed) · 0 unverified
centralized_gradient cifar10_gcsam_resnet50.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant fingerprinted no licence file found · pointer only · 27a3d56a09e0e7cf · report fixture: signature x:Tensor/4/float, use_gc:bool/0/bool, gc_conv_only:bool/0/bool · returned Tensor · level recorded 2026-09-22

Tasks

Fine-Grained Image ClassificationGeneral ClassificationImage Classificationimage-classification

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Batch NormalizationWeight Standardization

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