Deep networks prioritize easier examples over harder ones, leading to faster training.
problem Understanding how deep networks prioritize examples of varying difficulty.
method Investigated the effect of linear vs non-linear learning modes on example difficulty.
result Non-linear dynamics tend to sequentialize the learning of examples of increasing difficulty.
Paper shows training can improve GCN performance without changing architecture.
problem Training difficulty of GCNs limits their performance.
method Identified and mitigated energy loss during training.
result Significant decrease in training difficulties and notable performance boost.
Curriculum learning strategies improve GAN training speed and quality.
problem Training GANs is notoriously difficult and time-consuming.
method Proposes three curriculum learning strategies based on difficulty scores.
result Curriculum learning strategies lead to faster convergence and superior results.
Proposes a simple framework to balance task difficulty in multi-task learning.
problem Varying difficulty levels among different tasks in multi-task learning.
method Introduces a Balanced Multi-Task Learning (BMTL) framework that transforms training losses to balance task difficulty.
result Empirical studies show state-of-the-art performance of the proposed BMTL framework.
Time-continuous emotion prediction has become an increasingly compelling task in machine learning. Considerable efforts have been made to advance the performance of these systems. Nonetheless, the main focus has been the development of more sophisticated models and the incorporation of different expressive modalities (…
Wasserstein distance improves GANs by reducing training difficulties.
problem Training difficulties and arbitrary hyperparameters in GANs.
method Estimating Wasserstein distance for generative modeling.
result Various ways to estimate Wasserstein distance for generative models.
This work analyzes PINNs for advection-diffusion equations using NTK theory.
problem Understanding and resolving the training difficulties of PINNs for advection-diffusion equations.
method Neural Tangent Kernel (NTK) analysis of PINNs for the linear advection-diffusion equation (LAD).
result PINNs struggle due to spectral bias and convergence rate disparity, especially in advection-dominated and diffusion-dominated regimes.
The difficulty of classification affects the weight matrices' heavy tail appearance in deep learning networks.
problem Understanding the spectral properties of weight matrices in deep learning networks.
method Spectral analysis of weight matrices in different modules of DNNs, classification difficulty as a driving factor for heavy tail appearance.
result Higher classification difficulty leads to more frequent appearance of heavy tails in weight matrices spectra.
SAT improves adversarial training by smoothing the loss landscape through curriculum learning.
problem Adversarial training sacrifices clean accuracy for robustness and suffers from large generalization error.
method SAT uses curriculum learning to smooth the adversarial loss landscape, improving both clean and robust accuracy.
result SAT models improve clean and robust accuracy significantly compared to adversarial training and other baselines.
Reasoning models generate differently based on problem difficulty, not just length.
problem Understanding how reasoning models handle different problem difficulties.
method Examined hidden-state trajectories across competitive programming, mathematics, and Boolean satisfiability.
result Corrected trajectory geometry shows difficulty-dependent differences in reasoning models, with stronger effects in the code domain.
Entity Linking (EL) is the task of automatically identifying entity mentions in a piece of text and resolving them to a corresponding entity in a reference knowledge base like Wikipedia. There is a large number of EL tools available for different types of documents and domains, yet EL remains a challenging task where t…
The paper explores when to prioritize easy or hard samples in learning tasks.
problem Determining the optimal order of learning easy or hard samples.
method Theoretical analyses and experiments were conducted to propose and validate four priority modes.
result Four priority modes (easy-first, hard-first, medium-first, two-ends-first) can be flexibly applied.
We investigate the difficulties of training sparse neural networks and make new observations about optimization dynamics and the energy landscape within the sparse regime. Recent work of \citep{Gale2019, Liu2018} has shown that sparse ResNet-50 architectures trained on ImageNet-2012 dataset converge to solutions that a…
Learning shrinks hard tail, improving inference performance.
problem Improving inference performance in neural networks.
method Latent Instance Difficulty (LID) model analyzing fine-tuning of neural networks.
result Training-dependent inference scaling, with βexteff growing with sample size before saturating. New measure quantifies task difficulty for machine learning models.
problem Quantifying the inherent difficulty of machine learning tasks.
method Inductive bias complexity measure.
result Tasks requiring generalization over many dimensions are more difficult.
IAM improves deep RL by selectively storing influential past observations.
problem Training and performance issues with recurrent neural networks in high-dimensional environments.
method Influence-aware memory architecture that restricts input to variables influencing hidden state.
result IAM outperforms standard RNNs in training speed and policy performance.
Vanishing nodes cause hidden nodes to behave similarly, complicating deep neural network training.
problem Complicating deep neural network training due to hidden nodes behaving similarly.
method Introduced vanishing node indicator (VNI) to quantify node redundancy and correlated behavior.
result The effective number of nodes vanishes to one as VNI increases, indicating a challenging training scenario.
Gradient boosting with randomized trees reduces discontinuities and complexity.
problem Discontinuities in regression functions due to sparse training data.
method Gradient boosting machine with partially randomized decision trees.
result Improves robustness and computational efficiency of gradient boosting.
Adversarial training adds dynamic perturbations to neural networks for robustness.
problem Accuracy trade-off and lack of diversity in adversarial examples.
method Dynamic adversarial perturbations in the parameter space of neural networks, updating perturbation biases during training.
result Adversarial training with negligible cost and reduced accuracy trade-off.
Training neural networks is traditionally done by providing a sequence of random mini-batches sampled uniformly from the entire training data. In this work, we analyze the effect of curriculum learning, which involves the non-uniform sampling of mini-batches, on the training of deep networks, and specifically CNNs trai…
RB-Modulation trains free diffusion models without external adapters.
problem Training-free personalization of diffusion models with style and content control.
method Stochastic optimal control with a style descriptor and cross-attention aggregation.
result Precise content and style extraction and control without external adapters.
Study shows perceptual boost of visual attention varies with task difficulty and size.
problem Understanding how task-dependent the perceptual boost of visual attention is in natural settings.
method Designed and trained neural networks on various visual tasks, comparing results to baseline.
result Perceptual boost of attention is stronger with more difficult tasks and weaker with larger task sets.
Research uses CPS to estimate uncertainty in ML radio metric models.
problem Estimating uncertainty in machine learning models for radio metrics and path loss.
method Conformal Prediction (CP) in Conformal Predictive Systems (CPS) with diverse difficulty estimators.
result CPS models maintain high coverage and reliability across different cities.
RLVR training dynamics reveal an implicit curriculum that shapes learning progression.
problem Understanding how RLVR overcomes the long-horizon barrier.
method Developed a theory of training dynamics for RLVR on transformers, using Fourier analysis on finite groups.
result Mixed-difficulty training naturally follows an implicit curriculum, shaping the learning progression from easy to hard.
This work sets theoretical limits on meta-learning performance.
problem Understanding the difficulty of adapting machine learning models to real-world data distributions.
method Information-theoretic lower bounds on convergence rates for meta-learning algorithms.
result Theoretical bounds on parameter estimation error for hierarchical Bayesian models of meta-learning.
Logistic regression can handle noisy labels effectively when labels are imperfectly assigned by multiple experts.
problem Label noise in supervised classification due to manual labelling by multiple experts.
method Using approximate posterior probabilities of class membership from multiple experts to train logistic regression models.
result Logistic regression can be robust to label noise when classification difficulty is the only source of errors.
Paper provides conditions for reliable use of pre-trained embeddings in econometrics.
problem Uncertainty in using pre-trained embeddings for econometric tasks.
method Derives sufficient conditions and convergence rates for machine learning models with pre-trained embeddings.
result Establishes theoretical foundations for reliable use of pre-trained embeddings in econometrics.
sGPO optimizes RLVR training by balancing inference FLOPs and training efficiency.
problem RLVR training allocates rollout budget without considering query difficulty.
method sGPO trades inference FLOPs for reduced training FLOPs.
result sGPO matches or exceeds baseline performance while reducing training compute by a factor of three.
Single-stage neural architecture improves text-to-image synthesis.
problem Combining text and image generation is challenging.
method Used deep residual networks and sentence interpolation.
result Achieved state-of-the-art performance with single-stage training.
BatchNorm helps BNNs avoid exploding gradients during training.
problem Difficulty in training Binary Neural Networks (BNNs).
method Theoretical study and numerical experiments on the role of BatchNorm in BNNs.
result BatchNorm is crucial for avoiding exploding gradients in BNNs.
The impressive success of Generative Adversarial Networks (GANs) is often overshadowed by the difficulties in their training. Despite the continuous efforts and improvements, there are still open issues regarding their convergence properties. In this paper, we propose a simple training variation where suitable weights …
RLVR learning dynamics naturally create an implicit curriculum for transformers.
problem How rewards based on final outcomes help overcome the long-horizon barrier in reasoning models.
method Developed a theory of training dynamics for RLVR on transformers, using Fourier analysis on finite groups.
result Mixed-difficulty training induces an implicit curriculum that shapes the learning progression from easy to hard.
Symmetry in inverse problems leads to multiple solutions, but breaking symmetry helps deep learning.
problem Symmetry in physical systems causes multiple solutions in inverse problems, hindering deep learning.
method Careful symmetry breaking on training data helps solve inverse problems and improve deep learning performance.
result Symmetry breaking on training data significantly improves deep learning performance in inverse problems.
New model corrects bias in crowdsourced ratings for diverse items.
problem Bias and noise in crowdsourced ratings for training data.
method Bayesian rating model with item-level effects for difficulty, discriminativeness, and guessability.
result New model avoids bias in training data, improving model goodness of fit.
End-to-end framework learns from imperfect annotations directly.
problem Training machine learning models on imperfect human annotations.
method End-to-end framework merging aggregation with model training and modeling annotator competencies.
result Accuracy gains of up to 25% over state-of-the-art annotation aggregation methods.
Automatic estimation of relative difficulty of a pair of questions is an important and challenging problem in community question answering (CQA) services. There are limited studies which addressed this problem. Past studies mostly leveraged expertise of users answering the questions and barely considered other properti…
In the present work we propose a Deep Feed Forward network architecture which can be trained according to a sequential learning paradigm, where tasks of increasing difficulty are learned sequentially, yet avoiding catastrophic forgetting. The proposed architecture can re-use the features learned on previous tasks in a …
We investigate deep generative models that can exchange multiple modalities bi-directionally, e.g., generating images from corresponding texts and vice versa. A major approach to achieve this objective is to train a model that integrates all the information of different modalities into a joint representation and then t…
Dropout is a very effective way of regularizing neural networks. Stochastically "dropping out" units with a certain probability discourages over-specific co-adaptations of feature detectors, preventing overfitting and improving network generalization. Besides, Dropout can be interpreted as an approximate model aggregat…
Proposes a context-aware approach to deep autoencoder novelty detection.
problem Challenges of static distribution in novelty detection.
method Semi-supervised network architecture with auxiliary labels for contextual information.
result Single model achieves performance of individually trained models on various contexts.
Currently, deep neural networks are deployed on low-power portable devices by first training a full-precision model using powerful hardware, and then deriving a corresponding low-precision model for efficient inference on such systems. However, training models directly with coarsely quantized weights is a key step towa…
Neural network learns its size and structure during training.
problem Adapting neural network architecture to specific datasets.
method Flexible setup allowing neural network to learn size and topology during training.
result Trained networks achieve virtually identical performance and have learned optimal structure.
This work embeds annotations into a multidimensional space to measure classification difficulty.
problem Uncertainty in machine learning models during annotation phase.
method Develops a Bayesian Dirichlet-Multinomial framework to embed annotations and uses stochastic Expectation Maximization with MCMC.
result Embeddings reflect semantic similarities of original classes, aiding in measuring classification difficulty.
New model predicts video sequences with latent dynamics.
problem Predicting video sequences with inherent uncertainty.
method Introduces a novel stochastic temporal model with latent dynamics.
result Latent model outperforms prior state-of-the-art methods.
Data imbalance remains one of the most widespread problems affecting contemporary machine learning. The negative effect data imbalance can have on the traditional learning algorithms is most severe in combination with other dataset difficulty factors, such as small disjuncts, presence of outliers and insufficient numbe…
SimVAE trains interpretable VAEs using simulators.
problem Training interpretable generative models.
method Two-step process: train decoder to approximate simulator, then train encoder to invert it.
result Disentangled and interpretable latent space achieved.
The Resilient Propagation (Rprop) algorithm has been very popular for backpropagation training of multilayer feed-forward neural networks in various applications. The standard Rprop however encounters difficulties in the context of deep neural networks as typically happens with gradient-based learning algorithms. In th…
In this paper we propose two novel bounds for the log-likelihood based on Kullback-Leibler and the Rényi divergences, which can be used for variational inference and in particular for the training of Variational AutoEncoders. Our proposal is motivated by the difficulties encountered in training VAEs on continuous datas…