Papers › Noise Stability Optimization for Finding Flat Minima: A Hessian-based Regularization Approach

Noise Stability Optimization for Finding Flat Minima: A Hessian-based Regularization Approach

14 Jun 2023arXiv:2306.08553archive 2025-07-28

Hongyang R. Zhang, Dongyue Li, Haotian Ju

The training of over-parameterized neural networks has received much study in recent literature. An important consideration is the regularization of over-parameterized networks due to their highly nonconvex and nonlinear geometry. In this paper, we study noise injection algorithms, which can regularize the Hessian of the loss, leading to regions with flat loss surfaces. Specifically, by injecting isotropic Gaussian noise into the weight matrices of a neural network, we can obtain an approximately unbiased estimate of the trace of the Hessian. However, naively implementing the noise injection via adding noise to the weight matrices before backpropagation presents limited empirical improvements. To address this limitation, we design a two-point estimate of the Hessian penalty, which injects noise into the weight matrices along both positive and negative directions of the random noise. In particular, this two-point estimate eliminates the variance of the first-order Taylor's expansion term on the Hessian. We show a PAC-Bayes generalization bound that depends on the trace of the Hessian (and the radius of the weight space), which can be measured from data. We conduct a detailed experimental study to validate our approach and show that it can effectively regularize the Hessian and improve generalization. First, our algorithm can outperform prior approaches on sharpness-reduced training, delivering up to a 2.4% test accuracy increase for fine-tuning ResNets on six image classification datasets. Moreover, the trace of the Hessian reduces by 15.8%, and the largest eigenvalue is reduced by 9.7% with our approach. We also find that the regularization of the Hessian can be combined with weight decay and data augmentation, leading to stronger regularization. Second, our approach remains effective for improving generalization in pretraining multimodal CLIP models and chain-of-thought fine-tuning.

PaperPDFCodeCode 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 15 code samples harvested from 1 repository linked to this paper: 13 ran (of which 0 constructed an object rather than computing a result; 12 with no instrument failure: 0 honoured, 0 violated, 12 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified. Where Syntology's instrument failed: 1 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 15 samples from 1 repository, 13 ran (12 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.

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

virtuosoresearch/noise-stability-optimization officialmentioned in paperjaxMITSyntology: 15 samples, 13 ran (1 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

15 samples harvested: 13 ran (of which 0 constructed an object rather than computing a result; 12 with no instrument failure: 0 honoured, 0 violated, 12 with no contract checked; 1 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
12ran
2unverified

Licence: 0 of the 15 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from virtuosoresearch/noise-stability-optimization. “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.

virtuosoresearch/noise-stability-optimization official repository 15 samples, 13 ran (1 instrument-failed) · 2 unverified
compute_hessian_traces exps_on_graph_datasets/compute_noise_stability.py ran MIT (permissive) · 84dbf7cf0c1ad519 · report fixture: signature model:callable/0/?, loss:Tensor/0/float, maxIter:int/0/int, tol:float/0/float · returned ndarray · level recorded 2026-09-26
get_completion_data_path chain-of-thought-finetuning/src/paths.py ran MIT (permissive) · 083a2d17fc537dcd · report fixture: returned str · level recorded 2026-09-26
get_constant_schedule in 21 papers exps_on_text_datasets/transformers/optimization.py ran MIT (permissive) · 1141f8139411c789 · report fixture: signature optimizer:SGD/0/?, last_epoch:int/0/int · returned LambdaLR · level recorded 2026-09-26
get_constant_schedule_with_warmup in 25 papers exps_on_text_datasets/transformers/optimization.py ran MIT (permissive) · 8e243647722211b3 · report fixture: signature optimizer:SGD/0/?, num_warmup_steps:int/0/int, last_epoch:int/0/int · returned LambdaLR · level recorded 2026-09-26
get_dataset_path in 10 papers chain-of-thought-finetuning/src/paths.py ran MIT (permissive) · 3df12d92fd1ba3ec · report fixture: returned str · level recorded 2026-09-26
get_linear_schedule_with_warmup in 39 papers exps_on_text_datasets/transformers/optimization.py ran · our draft was wrong MIT (permissive) · afba2c06a42074fe · report fixture: signature optimizer:SGD/0/?, num_warmup_steps:int/0/int, num_training_steps:int/0/int, last_epoch:int/0/int · returned LambdaLR · level recorded 2026-09-26
get_split_path chain-of-thought-finetuning/src/paths.py ran MIT (permissive) · c2bc595db0df3786 · report fixture: returned str · level recorded 2026-09-26
get_trainable_parameters chain-of-thought-finetuning/compute_hessian_traces.py ran MIT (permissive) · 51a2376d3165e058 · report fixture: signature model:callable/0/? · returned generator · level recorded 2026-09-26
get_weights in 34 papers exps_on_graph_datasets/compute_noise_stability.py ran MIT (permissive) · 7f5a965a9547d732 · report fixture: signature model:callable/0/? · returned list · level recorded 2026-09-26
label_noise exps_on_image_datasets/data_loader/random_noise.py ran MIT (permissive) · 485783150c7893b6 · report fixture: signature dataset:SimpleDataset/0/?, train_indices:ndarray/1/int, noise_rate:float/0/float, symmetric:bool/0/bool · returned tuple · level recorded 2026-09-26
noisy_labeler exps_on_image_datasets/data_loader/random_noise.py ran MIT (permissive) · 430d1fb772405f2e · report fixture: signature train_data_loader:list/1/tuple, dataset:SimpleDataset/0/?, train_indices:list/1/int, aux_model:callable/0/?, device:str/0/str · returned tuple · level recorded 2026-09-26
perturbe_model_weights exps_on_image_datasets/compute_noise_stability.py ran MIT (permissive) · aab5cc3f4ce427b5 · report fixture: signature state_dict:dict/0/?, eps:float/0/float · returned dict · level recorded 2026-09-26
perturbe_model_weights exps_on_text_datasets/compute_noise_stability.py ran MIT (permissive) · 2c8622034db2b781 · report fixture: signature state_dict:dict/0/?, eps:float/0/float, use_neg:bool/0/bool, perturbation:dict/0/? · returned tuple · level recorded 2026-09-26
create_optimizer in 37 papers exps_on_text_datasets/transformers/optimization_tf.py unverified MIT (permissive) · 6c2c33724affff7b · report

Tasks

Data AugmentationImage ClassificationStochastic Optimizationimage-classification

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Average PoolingConvolutionGlobal Average PoolingKaiming InitializationMax PoolingSGDSharpness-Aware MinimizationWeight Decay

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