Enhances learning of structured distributions using nonlinear denoising score matching.
problem Learning structured distributions from noisy data.
method Latent Nonlinear Denoising Score Matching (LNDSM) integrating nonlinear dynamics with VAE-based latent score matching.
result LNDSM achieves superior sample quality and variability compared to structure-agnostic methods.
PDNS tackles multimodal sampling challenges using proximal point method.
problem Multimodal distributions with significant barriers between modes.
method Proximal point method on path measures, decomposing into simpler subproblems.
result PDNS effectively promotes thorough exploration across modes.
Recently, the deep-belief-networks (DBN) based voice activity detection (VAD) has been proposed. It is powerful in fusing the advantages of multiple features, and achieves the state-of-the-art performance. However, the deep layers of the DBN-based VAD do not show an apparent superiority to the shallower layers. In this…
Denoising autoencoders (DAEs) are powerful deep learning models used for feature extraction, data generation and network pre-training. DAEs consist of an encoder and decoder which may be trained simultaneously to minimise a loss (function) between an input and the reconstruction of a corrupted version of the input. The…
UDM reparameterization improves language model generation.
problem Mismatch between UDM training objective and denoising posterior.
method Leave-one-out denoising and absorbing state reformulation.
result Improved UDM generation through leave-one-out parameterization.
G-SimCLR improves unsupervised learning by clustering images into pseudo labels.
problem Improving unsupervised learning for image recognition.
method Proposes a method to cluster images into pseudo labels to batch images of the same category.
result Comparable performance enhancements on CIFAR10 and ImageNet datasets.
Conservation laws improve diffusion model training by optimizing likelihood.
problem Training diffusion models with denoising objectives.
method Developed conservation laws based on GEXIT functions for memoryless noise processes.
result Unified characterization of diffusion model likelihood, reducing training to learning marginal posteriors.
Quantum machine learning uses quantum cross entropy to minimize loss, but measurement loss affects this process.
problem Quantum machine learning's loss minimization through cross entropy is affected by measurement outcomes.
method Defined quantum cross entropy, proved its lower bounds, and investigated its relation to quantum fidelity and likelihood.
result Quantum cross entropy is lower-bounded by negative log-likelihood when derived from quantum data, but measurement outcomes can cause loss.
SJDs unify masked, continuous, and hybrid diffusion models.
problem Unified modeling of diffusion processes.
method Continuous-time Markov processes with token embeddings and hazard rates.
result Unified model recovers masked, continuous, and hybrid diffusion as limits.
Cross-entropy loss linked to metric learning, outperforming complex pairwise losses.
problem Improving metric learning performance without complex optimization schemes.
method Theoretical analysis linking cross-entropy to pairwise losses, showing cross-entropy as an upper bound and equivalent to mutual information maximization.
result Minimizing cross-entropy is equivalent to maximizing mutual information, leading to state-of-the-art performance.
Unified interpretation of softmax cross-entropy and negative sampling for knowledge graph embedding.
problem Lack of theoretical relationship between softmax cross-entropy and negative sampling loss functions in knowledge graph embedding.
method Used Bregman divergence to provide a unified interpretation of the two loss functions.
result Theoretical findings for fair comparison of softmax cross-entropy and negative sampling are derived.
PAC-Bayesian bounds for MLPs with cross entropy loss validated.
problem Generalization bounds for MLPs with cross entropy loss.
method Introduced probabilistic explanations and proved PAC-Bayesian bounds using ELBO.
result MLPs with cross entropy loss inherently guarantee PAC-Bayesian generalization bounds.
New method detects changes by maximizing cross-entropy, outperforming existing techniques.
problem Detecting abrupt changes in data streams without labeled examples.
method Maximizes cross-entropy between segments to find change points, using dynamic programming.
result Outperforms three state-of-the-art approaches on challenging datasets.
We propose a max-pooling based loss function for training Long Short-Term Memory (LSTM) networks for small-footprint keyword spotting (KWS), with low CPU, memory, and latency requirements. The max-pooling loss training can be further guided by initializing with a cross-entropy loss trained network. A posterior smoothin…
Square loss performs comparably or better than cross-entropy in neural architectures for various tasks.
problem The superiority of cross-entropy loss over square loss in classification tasks is debated.
method Comparison of several neural architectures on NLP, ASR, and computer vision datasets using both loss functions.
result Square loss often produces better results in the majority of tasks, especially in NLP and ASR.
The paper explores how the depth of neural networks affects their ability to represent data accurately.
problem Understanding the implicit bias and rank of neural networks with large depth.
method Analyzing the convergence of representation cost to a notion of rank as network depth increases, and investigating conditions for recovering the true rank of data.
result There is a range of network depths where the true rank of data is recovered, and this affects the topology of class boundaries.
Theoretical analysis of cross-entropy loss functions and their robustness.
problem Guarantees for using cross-entropy as a surrogate loss function.
method Theoretical analysis of a broad family of loss functions, including cross-entropy.
result First H-consistency bounds for comp-sum losses and smooth adversarial comp-sum losses. Proposes MGCE for improved classification performance.
problem Optimizing between robustness and optimization difficulty in classification.
method Minimax formulation of GCE leading to convex optimization over margins.
result MGCE achieves strong accuracy and better calibration, especially in noisy labels.
New neural network approach using mutual information.
problem Training neural networks for imbalanced datasets.
method Converts neural network classifiers to mutual information evaluators.
result New form of softmax leads to better classification accuracy, especially for imbalanced datasets.
Study Transformer layers under cross-entropy training using mean field control.
problem Understanding the behavior of Transformer layers in cross-entropy training.
method Continuous-depth mean field control analysis, treating depth as time and layer parameters as controls.
result Derivation of a Pontryagin condition for the limiting population problem, involving the softmax residual.
This paper proposes a set of new error criteria and learning approaches, Adaptive Normalized Risk-Averting Training (ANRAT), to attack the non-convex optimization problem in training deep neural networks (DNNs). Theoretically, we demonstrate its effectiveness on global and local convexity lower-bounded by the standard …
In deep neural network, the cross-entropy loss function is commonly used for classification. Minimizing cross-entropy is equivalent to maximizing likelihood under assumptions of uniform feature and class distributions. It belongs to generative training criteria which does not directly discriminate correct class from co…
Paper compares optimal denoising methods for generative models, finding different results based on data regularity.
problem Optimizing denoising in score-based generative models for various data types.
method Comparison of full-denoising and half-denoising approaches, analyzing performance in terms of distribution distances.
result Different denoising methods perform better under different data regularity conditions.
Proposes a method to solve deep neural networks' local minimum problem.
problem Local minimum problem in deep neural networks training.
method Transforms cross-entropy loss into risk-averse error criterion, adjusts RSI, and uses convexity region.
result Trained deep learning machine is expected to be inside a global minimum's attraction basin.
Paper proposes a new loss function for conditional models using soft targets.
problem Improving generalization performance of deep neural networks on supervised classification tasks.
method Introduces a new loss function compatible with soft targets, based on noise contrastive estimation.
result Soft target InfoNCE loss performs on par with cross-entropy baselines and outperforms other losses.
Convolution Neural Networks (CNN) have recently achieved state-of-the art performance on handwritten Chinese character recognition (HCCR). However, most of CNN models employ the SoftMax activation function and minimize cross entropy loss, which may cause loss of inter-class information. To cope with this problem, we pr…
Image super-resolution (SR) is an underdetermined inverse problem, where a large number of plausible high-resolution images can explain the same downsampled image. Most current single image SR methods use empirical risk minimisation, often with a pixel-wise mean squared error (MSE) loss. However, the outputs from such …
Paper proposes AXE loss for non-autoregressive machine translation, improving performance.
problem Challenges in training non-autoregressive models due to lack of autoregressive factors and cross entropy loss penalties.
method Proposes aligned cross entropy (AXE) loss function using a differentiable dynamic program for better word order alignment.
result AXE-based training improves performance on major WMT benchmarks and sets a new state of the art for non-autoregressive models.
Proposes squentropy loss for improved classification accuracy and model calibration.
problem Theoretical and empirical evidence for cross-entropy loss is lacking.
method Introduces squentropy loss as the sum of cross-entropy and average square loss over incorrect classes.
result Squentropy loss outperforms cross-entropy and rescaled square losses in classification accuracy and model calibration.
Unified method for simultaneous denoising and clustering.
problem Clustering noisy signals.
method Sparse convex wavelet clustering with fusion and group-sparse penalties.
result Unified approach that denoises and clusters simultaneously.
DDPD separates generation into planning and denoising for improved efficiency.
problem Efficiently denoise corrupted data during generation.
method Separates generation into a planner and denoiser, selecting denoising positions based on corruption severity.
result DDPD outperforms traditional methods on language and image generation benchmarks.
CADO optimizes heatmap-based solvers for cost minimization, overcoming performance limitations.
problem Heatmap-based solvers lack objective alignment for cost minimization.
method CADO uses Reinforcement Learning to optimize solution cost directly, introducing Label-Centered Reward and Hybrid Fine-Tuning.
result CADO achieves state-of-the-art performance across diverse benchmarks.
Improved self-supervised denoising for Poisson-Gaussian noise.
problem Handling Poisson-Gaussian noise in self-supervised denoising.
method Extended blindspot model, improved training scheme without hyperparameters.
result Improved denoising performance on microscope image benchmarks.
We consider the problem of estimating a low-rank matrix from a noisy observed matrix. Previous work has shown that the optimal method depends crucially on the choice of loss function. In this paper, we use a family of weighted loss functions, which arise naturally for problems such as submatrix denoising, denoising wit…
Image denoising is an important pre-processing step in medical image analysis. Different algorithms have been proposed in past three decades with varying denoising performances. More recently, having outperformed all conventional methods, deep learning based models have shown a great promise. These methods are however …
Improved language models using ratio-matching and KL divergence.
problem Efficiently modeling discrete data with diffusion models.
method Introduced new theorems and a novel CTMC transition-rate matrix for ratio-matching and KL divergence.
result 10-15% improvement in perplexity and faster training steps.
Improved CEM for fast real-time planning in high-dimensional control tasks.
problem Sampling inefficiency of CEM in real-time planning.
method Novel additions to CEM including temporally-correlated actions and memory.
result 2.7-22x less samples and 1.2-10x performance increase.
Gen-CUDE is a neural network for denoising noisy channels.
problem Denoising in finite-input, general-output noisy channels.
method Unsupervised neural network trained on noisy data.
result Gen-CUDE achieves better denoising results than other methods.
We present the Tamed Cross Entropy (TCE) loss function, a robust derivative of the standard Cross Entropy (CE) loss used in deep learning for classification tasks. However, unlike other robust losses, the TCE loss is designed to exhibit the same training properties than the CE loss in noiseless scenarios. Therefore, th…
GDiff tackles blind denoising with Gibbs sampling and Monte Carlo inference.
problem Blind denoising of signals with unknown noise parameters.
method Gibbs Diffusion (GDiff) method that alternates sampling steps from a conditional diffusion model and a Monte Carlo sampler.
result GDiff achieves blind denoising of natural images and cosmic microwave background data.
While it is believed that denoising is not always necessary in many big data applications, we show in this paper that denoising is helpful in urban traffic analysis by applying the method of bounded total variation denoising to the urban road traffic prediction and clustering problem. We propose two easy-to-implement m…
Generative Cross-Entropy improves classification with fewer labels.
problem Limited sample efficiency of cross-entropy loss in data-scarce scenarios.
method Proposes Generative Cross-Entropy (GenCE), a new loss function that incorporates generative principles into a standard discriminative network.
result Generative Cross-Entropy outperforms traditional cross-entropy loss across various datasets and conditions.
This paper tackles denoising of complex measures using optimal transport and curvature analysis.
problem Denoising of complex, possibly non-log-concave measures.
method Score function and optimal transport theory to revert Langevin diffusion chains.
result The difficulty of denoising depends on the curvature complexity of the initial measure at specific SNR scales.
New denoisers improve signal recovery from noisy data without knowing noise distribution.
problem Denoising signals when only noise level is known, not distribution.
method Universal denoisers that shrink PY toward PX with higher-order accuracy. result Achieves O(σ4) and O(σ6) accuracy in matching generalized moments and densities. New model reduces sampling cost in diffusion models, making them faster and applicable to real-world applications.
problem Challenges in generating high-quality samples, mode coverage, and fast sampling in deep generative models.
method Proposes denoising diffusion GANs that model each denoising step using a multimodal conditional GAN to reduce sampling cost.
result Demonstrates 2000imes faster sampling on CIFAR-10 dataset while maintaining competitive sample quality and diversity. Nyström approximation for scalable operator learning
problem Scalability of operator learning for large datasets
method Nyström subsampling with operator learning
result Minimax-optimal convergence rates for functional outputs
UDVD uses deep learning to denoise videos without supervision.
problem Lack of clean video data for training deep learning models.
method UDVD is a CNN trained solely on noisy video data, adapting to local motion.
result UDVD performs as well as supervised methods, even with limited training data.
Unified framework for denoising models across various spaces.
problem Improving generative models and approximate posterior simulation.
method Generalizing denoising diffusions to a broader class of spaces using a new extension of score matching.
result Unified models for denoising and posterior simulation.