arXiv research

Two-layer ReLU networks can overfit without harm, study finds.

Study shows flipping a small subset of labels can severely damage machine learning models.

Deep Partition Aggregation defends against poisoning attacks with provable certificates.

New method makes machine learning models robust to label flipping attacks.

Many machine learning systems rely on data collected in the wild from untrusted sources, exposing the learning algorithms to data poisoning. Attackers can inject malicious data in the training dataset to subvert the learning process, compromising the performance of the algorithm producing errors in a targeted or an ind…

In label-noise learning, \textit{noise transition matrix}, denoting the probabilities that clean labels flip into noisy labels, plays a central role in building \textit{statistically consistent classifiers}. Existing theories have shown that the transition matrix can be learned by exploiting \textit{anchor points} (i.e…

Label manipulation attacks are a subclass of data poisoning attacks in adversarial machine learning used against different applications, such as malware detection. These types of attacks represent a serious threat to detection systems in environments having high noise rate or uncertainty, such as complex networks and I…

Ridge regression shows different behaviors in binary classification with noisy labels.

Efficiently poisons offline RLHF models by flipping preference labels.

Novel defense algorithm improves SVMs against data poisoning attacks.

Transformers trained on random classification tasks generalize well and can overfit without error.

New algorithm uses conditionally invariant components to improve domain adaptation performance.

Federated learning has a variety of applications in multiple domains by utilizing private training data stored on different devices. However, the aggregation process in federated learning is highly vulnerable to adversarial attacks so that the global model may behave abnormally under attacks. To tackle this challenge, …

We introduce a tunable loss function called $α$-loss, parameterized by $α\in (0,\infty]$, which interpolates between the exponential loss ($α= 1/2$), the log-loss ($α= 1$), and the 0-1 loss ($α= \infty$), for the machine learning setting of classification. Theoretically, we illustrate a fundamental connection between $…

In this paper, we proposed a general framework for data poisoning attacks to graph-based semi-supervised learning (G-SSL). In this framework, we first unify different tasks, goals, and constraints into a single formula for data poisoning attack in G-SSL, then we propose two specialized algorithms to efficiently solve t…

Machine learning (ML) over distributed multi-party data is required for a variety of domains. Existing approaches, such as federated learning, collect the outputs computed by a group of devices at a central aggregator and run iterative algorithms to train a globally shared model. Unfortunately, such approaches are susc…

This work is motivated by the problem of image mis-registration in remote sensing and we are interested in determining the resulting loss in the accuracy of pattern classification. A statistical formulation is given where we propose to use data contamination to model and understand the phenomenon of image mis-registrat…

A robust loss for anomaly mitigation and unsupervised contamination classification

Nowadays, how to effectively evaluate visual properties has become a popular topic for fine-grained visual comprehension. In this paper we study the problem of how to estimate such visual properties from a ranking perspective with the help of the annotators from online crowdsourcing platforms. The main challenges of ou…

Adaptive sampler improves recommendation for implicit feedback data.

Measures three types of noise in LLM evaluations.

The study investigates noise effects on parameter estimation for Ornstein-Uhlenbeck processes.

Noise in SGD affects overparameterized models, favoring sparse solutions.

L2R learns to denoise images without needing noise distribution knowledge.

Interpolating label noise makes models vulnerable to adversarial attacks.

Generative adversarial networks (GANs) are neural networks that learn data distributions through adversarial training. In intensive studies, recent GANs have shown promising results for reproducing training images. However, in spite of noise, they reproduce images with fidelity. As an alternative, we propose a novel fa…

We present a noise-injected version of the Expectation-Maximization (EM) algorithm: the Noisy Expectation Maximization (NEM) algorithm. The NEM algorithm uses noise to speed up the convergence of the EM algorithm. The NEM theorem shows that injected noise speeds up the average convergence of the EM algorithm to a local…

BeGIN benchmarks GNNs for instance-dependent label noise in graphs.

Enhanced consistency bounds derived for classification under a new noise condition.

Proposes a progressive label correction method for feature-dependent label noise.

New methods for handling time-varying label noise in time series classification.

NoiseRank reduces label noise without supervision, improving classification accuracy.

Paper tackles online control of linear systems with unbounded noise.

Efficient algorithm for halfspaces with specific noise conditions.

Study on noise models for noisy labels in NLP.

Using a Bayesian approach, we consider the problem of recovering sparse signals under additive sparse and dense noise. Typically, sparse noise models outliers, impulse bursts or data loss. To handle sparse noise, existing methods simultaneously estimate the sparse signal of interest and the sparse noise of no interest.…

TrustNet robustly learns noise patterns from trusted data to improve weakly-supervised classification.

Noise enhancement improves generalization in training.

HeMPPCAT improves PCA for data with varying noise.

The paper examines various RL algorithms to address overestimation and noise issues.

Geometry-aware noise improves model generalization on complex manifolds.

A method uses confidence scores to handle noisy labels for each instance.

New insights into noise distribution for self-supervised learning.

New method designs joint initial noises for diffusion models to improve diversity and alignment.

Paper improves SLCB regret bound for bounded noise.

Proposes DISCO, the first CVI for density-based clustering with noise.

We address noisy Euclidean distances in high dimensions, estimating noise levels and correcting distances.

DA-GNN improves robustness of GNNs by modeling noise dependencies.