Label smoothing improves model performance even with noisy labels.
problem Mitigating label noise in deep learning models.
method Examined label smoothing as a technique to cope with label noise and compared it to loss-correction methods.
result Label smoothing is competitive with loss-correction techniques under label noise and beneficial for distillation from noisy data.
Label smoothing improves generalization by controlling generalization loss.
problem Lack of mathematical understanding of label smoothing's effectiveness.
method Proposed a theoretical framework to show how label smoothing controls generalization loss in the label noise setting.
result Predicted an optimal label smoothing point that minimizes generalization loss.
Label smoothing improves model calibration and generalization but harms distillation.
problem Understanding the effects of label smoothing on model calibration and distillation.
method Empirical evaluation and visualization of network representations.
result Label smoothing improves model calibration but harms knowledge distillation.
Label-Smoothing improves adversarial robustness of deep learning models.
problem Improving the robustness of deep learning models against adversarial attacks.
method Label-Smoothing, including adversarial, Boltzmann, and second-best variants, applied to various models and datasets.
result Label-Smoothing enhances adversarial robustness across multiple attacks and datasets.
Improved OOD detection using label smoothing and k-NN density estimates.
problem Detecting out-of-distribution examples in classification models.
method Label smoothing and k-NN density estimate on intermediate activations.
result Label smoothing improves OOD detection performance, both theoretically and empirically.
Paper introduces SLS to improve label smoothing regularization.
problem Improve generalization of neural networks by softening labels.
method Structural Label Smoothing (SLS) to mitigate bias in Bayes error rate.
result Empirical validation shows SLS outperforms traditional label smoothing.
Enhances deep networks robustness with data mollification and label smoothing.
problem Improving deep neural networks' robustness against corruptions.
method Coupling data mollification (image noising and blurring) with label smoothing.
result Improved robustness and uncertainty quantification on corrupted image benchmarks.
In this article, we mathematically study several GAN related topics, including Inception score, label smoothing, gradient vanishing and the -log(D(x)) alternative. --- An advanced version is included in arXiv:1703.02000 "Activation Maximization Generative Adversarial Nets". Please refer Section 6 in 1703.02000 for deta…
Pairwise Label Smoothing improves deep model generalization by reducing overconfidence.
problem Improving deep model generalization through regularization.
method PLS smooths labels for pairs of samples, learning distribution mass during training.
result PLS significantly outperforms LS and baseline models, reducing up to 30% classification error.
Study analyzes label smoothing in deep learning optimization.
problem Understanding label smoothing's impact on deep learning optimization.
method Analysis of stochastic gradient descent with label smoothing for non-convex problems.
result Label smoothing can speed up convergence by reducing variance.
Self-distillation improves model performance by increasing teacher diversity and smoothing predictions.
problem Improving model generalization and performance through self-distillation.
method Interpreting self-distillation as MAP estimation and proposing instance-specific label smoothing.
result Self-distillation enhances model performance by increasing teacher diversity and smoothing predictions.
KGNN-LS improves recommender systems using knowledge graphs and label smoothness.
problem Improving recommender systems through better user-item embeddings.
method KGNN-LS combines knowledge graphs, user-specific embeddings, and label smoothness regularization.
result KGNN-LS outperforms state-of-the-art baselines and handles cold-start scenarios.
The study examines label smoothing to improve confidence calibration in fine-tuned LLMs.
problem Improving confidence calibration in fine-tuned large language models (LLMs) after instruction tuning.
method Examine various open-sourced LLMs, label smoothing, and custom kernel design.
result Label smoothing is effective in maintaining confidence calibration but faces challenges in large vocabulary LLMs.
Improves confidence calibration in neural networks by smoothing labels based on class similarity.
problem Improving confidence calibration in deep neural networks for safety-critical applications.
method Proposes a novel label smoothing technique where label values are based on similarities with the reference class, using different similarity measurements.
result Consistently outperforms state-of-the-art calibration techniques on various datasets and network architectures.
Simple regularization methods mimic adversarial training's robustness.
problem Expensive adversarial training for robustness.
method Label smoothing and logit squeezing.
result Achieves strong adversarial robustness without adversarial examples.
Label smoothing improves model robustness against misspecification.
problem Improving model robustness against model misspecification.
method Introducing modified label smoothing (MLSLR) that maintains consistent probability estimation while modifying the loss function.
result MLSLR exhibits higher robustness against model misspecification than conventional label smoothing.
Logit regularization induces logit clustering, affecting classifier performance.
problem Understanding the mechanism of logit regularization in classification.
method Analysis of logit regularization in linear classification, proving logit clustering leads to Fisher's Linear Discriminant alignment.
result Logit regularization can halve critical sample complexity and induce robust generalization.
CutMix training technique improves spatial locality in Vision Transformers.
problem Improving spatial locality in Vision Transformers trained from scratch.
method Comparison of Baseline and Modern training protocols on CIFAR-10, CIFAR-100, and Tiny-ImageNet.
result CutMix training component significantly reduces Mean Attention Distance (MAD) in early layers of Vision Transformers.
Proposes LsrKD and MrKD to improve neural network training performance.
problem Improving neural network training performance, especially on deep networks.
method Extends Label Smoothing Regularization with Self-Knowledge Distillation, introducing LsrKD and MrKD.
result LsrKD and MrKD significantly improve training performance on deep neural networks.
Researchers propose better probabilistic models for deep learning.
problem Using cross-entropy loss for non-categorical data.
method Introducing continuous-categorical distribution and proposing probabilistic alternatives.
result Potential for outperformance in deep learning models with proper probabilistic treatment.
Method reweights instances and classes to improve robustness in noisy data.
problem Improving deep learning performance in the presence of label noise.
method Formulates constrained optimization problems to assign importance weights to instances and class labels.
result Significant performance gains observed in benchmark datasets with label noise.
A new unsupervised contrastive learning framework improves time series representation learning.
problem Lack of labeled data in time series data.
method Proposes an unsupervised contrastive learning framework using a novel contrastive loss and data augmentation.
result Framework outperforms other approaches on univariate and multivariate time series, and benefits transfer learning.
Improves model calibration by addressing adversarially unrobust inputs.
problem Miscalibrated predictions and lack of adversarial robustness in neural networks.
method AR-AdaLS, an adaptive label smoothing method that considers adversarial robustness.
result AR-AdaLS improves model calibration, even under distributional shifts.
This paper improves understanding of knowledge distillation and its effects.
problem Improving model quality with a fixed capacity budget.
method Categorizing teacher's knowledge into three levels and studying their effects on distillation.
result The three hierarchical levels of teacher's knowledge (universe, domain, instance) significantly impact knowledge distillation.
Simple methods boost sound event classifier accuracy by 2.5%.
problem Label noise in sound event classification datasets.
method Label smoothing regularization, mixup, and noise-robust loss functions.
result Simple methods can boost accuracy by up to 2.5%.
Mixup improves model accuracy and calibration through data transformation and random perturbation.
problem Improving model accuracy and calibration in machine learning.
method Interprets Mixup as empirical risk minimization with data transformation and random perturbation.
result Mixup induces multiple known regularization schemes that prevent overfitting and overconfident predictions.
Study shows multilingual LLM calibration effects improve model confidence but not accuracy.
problem Improving multilingual language model calibration in low-resource settings.
method Analysis of two multilingual benchmarks using instruction-tuning and label smoothing.
result Model confidence increases in low-resource languages after instruction-tuning but accuracy improvements are marginal.
Mixup training improves deep neural network calibration and predictive uncertainty.
problem Improving the reliability and confidence of deep neural network predictions.
method Mixup training combines random pairs of images and labels during training.
result DNNs trained with mixup show significantly better calibration and predictive uncertainty.
Data augmentation is usually used by supervised learning approaches for offline writer identification, but such approaches require extra training data and potentially lead to overfitting errors. In this study, a semi-supervised feature learning pipeline was proposed to improve the performance of writer identification b…
This paper describes FBK's submission to the end-to-end English-German speech translation task at IWSLT 2018. Our system relies on a state-of-the-art model based on LSTMs and CNNs, where the CNNs are used to reduce the temporal dimension of the audio input, which is in general much higher than machine translation input…
Self-teaching networks improve deep neural networks' generalization.
problem Improving deep neural networks' generalization capacity.
method Generates soft supervision labels to train lower layers, using an auxiliary loss to mimic the output layer.
result Self-teaching network achieves consistent improvements and outperforms existing methods in speech recognition tasks.
New method observes learning paths to improve model supervision.
problem Improving model performance through better supervision.
method Observing learning paths to refine labels and propose Filter-KD.
result Models can refine bad labels through a 'zig-zag' learning path.
Adaptive label regularization improves neural network performance.
problem Improving deep neural network performance through label regularization.
method Proposes an adaptive label regularization method that updates label representation online.
result Significant improvement in image recognition and text classification tasks.
Evolutionary methods improve neural network loss functions, reducing overfitting.
problem Improving neural network performance and preventing overfitting.
method Evolutionary computation to optimize loss functions, balancing error pull and overfitting push.
result Evolved loss functions effectively reduce overfitting, leading to better performance and robustness.
Domain adaptation is transfer learning which aims to generalize a learning model across training and testing data with different distributions. Most previous research tackle this problem in seeking a shared feature representation between source and target domains while reducing the mismatch of their data distributions.…
Attention-based encoder-decoder architectures such as Listen, Attend, and Spell (LAS), subsume the acoustic, pronunciation and language model components of a traditional automatic speech recognition (ASR) system into a single neural network. In previous work, we have shown that such architectures are comparable to stat…
Regularization may not fully protect against membership inference attacks.
problem Mitigating membership inference attacks in deep learning models.
method Evaluation of 8 regularization mechanisms on three image classification tasks.
result Certain regularization mechanisms can inadvertently help membership inference attacks.
Practically, we are often in the dilemma that the labeled data at hand are inadequate to train a reliable classifier, and more seriously, some of these labeled data may be mistakenly labeled due to the various human factors. Therefore, this paper proposes a novel semi-supervised learning paradigm that can handle both l…
Unified model combines GCN and LPA for better node classification.
problem Combining GCN and LPA for improved node classification.
method Unified model that unifies GCN and LPA, learns edge weights and attention weights.
result Unified model outperforms state-of-the-art GCN-based methods in node classification accuracy.
In applications of supervised learning applied to medical image segmentation, the need for large amounts of labeled data typically goes unquestioned. In particular, in the case of brain anatomy segmentation, hundreds or thousands of weakly-labeled volumes are often used as training data. In this paper, we first observe…
This paper improves anomaly detection in lane rendering images for safer navigation.
problem Anomalies in lane rendering images can mislead drivers, posing safety risks.
method Proposes a four-phase pipeline using Transformer models, self-supervised pre-training, and fine-tuning.
result The pipeline enhances detection accuracy and reduces training time.
This paper examines how different loss functions affect neural network features and performance.
problem Investigating which loss function is best for deep neural networks.
method Examining last-layer features of deep networks and drawing inspiration from the Neural Collapse phenomenon.
result All relevant loss functions (CE, LS, FL, MSE) produce equivalent features and similar performance.