Proposes neuron alignment to optimize mode connectivity in neural networks.
problem Understanding and optimizing mode connectivity in deep neural networks.
method Introduces neuron alignment to approximate optimal weight permutations and improve mode connectivity.
result Neuron alignment significantly alleviates robust loss barriers and improves model robustness and accuracy.
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.
LPL optimizes embeddings to align local neighborhoods, improving cross-lingual word alignment.
problem Aligning embeddings across different datasets and languages.
method Locality Preserving Loss (LPL) optimizes model to project embeddings while maintaining local neighborhoods and aligning them.
result LPL-based alignment leads to better and consistent accuracy, especially in small training set settings.
Adaptive loss function improves performance by aligning training and evaluation metrics.
problem Loss-metric mismatch in machine learning training.
method Adaptive loss alignment through meta-learning of a dynamic loss function.
result Significant performance improvements across various tasks and data.
Extends reinforcement learning alignment to scalar rewards, improving math reasoning.
problem Designing reinforcement learning algorithms for general LLM alignment.
method Introduces f-GRPO and f-HAL, estimating f-divergences between reward-aligned and unaligned distributions.
result Improves math reasoning RLVR tasks and mitigates reward hacking.
SGD updates align with a low-rank subspace but do not lead to further loss reduction.
problem Understanding the training dynamics of deep neural networks, particularly the role of the dominant subspace.
method Exploring whether neural networks can be trained within the dominant subspace of the loss Hessian.
result SGD updates, when projected onto the dominant subspace, do not decrease the training loss further, suggesting spurious alignment.
The goal of temporal alignment is to establish time correspondence between two sequences, which has many applications in a variety of areas such as speech processing, bioinformatics, computer vision, and computer graphics. In this paper, we propose a novel temporal alignment method called least-squares dynamic time war…
This work optimizes alignment and uniformity of features on a hypersphere for better downstream performance.
problem Improving the performance of contrastive representation learning.
method Identifying and optimizing alignment and uniformity of features on a hypersphere.
result Directly optimizing alignment and uniformity leads to comparable or better performance than contrastive learning.
DFA trains deep networks by aligning weights then memorizing data.
problem Understanding why DFA works for some networks but not others.
method Two-step learning process: alignment followed by memorization.
result DFA aligns weights to maximize gradient alignment, breaking degeneracy.
New DG method minimizes barycentric alignment and reconstruction loss.
problem Improving domain generalization in machine learning.
method Introduces a new upper bound and WBAE algorithm.
result WBAE outperforms state-of-the-art DG algorithms.
We consider neural network training, in applications in which there are many possible classes, but at test-time, the task is a binary classification task of determining whether the given example belongs to a specific class, where the class of interest can be different each time the classifier is applied. For instance, …
Proposes a method to align language and image data.
problem Aligning language and image data for better understanding.
method Uses triplet loss to learn consistent embeddings of language and images.
result Outperforms four baselines across multiple metrics.
EAST aligns neural network classifiers with user-defined evaluation metrics.
problem Mismatch between neural network training and evaluation metrics leads to suboptimal performance.
method EAST uses dynamic thresholding, soft-set confusion matrix, and annealing to align neural network predictions with target evaluation metrics.
result EAST improves alignment between training objectives and evaluation metrics, outperforming existing methods.
Mitigates overoptimization in RLHF by reformulating SFT loss as a preference optimization loss.
problem Overoptimization in RLHF where reward model misguides generative model.
method Proposes a theoretical algorithm that minimizes maximum likelihood estimation and reward penalty, reformulates as simple objective combining preference optimization and supervised learning losses.
result Improved performance of RPO compared to DPO baselines in aligning LLMs.
SFAVEL distills features from language models for fact verification without annotations.
problem Fact verification with semantically meaningful and compact features.
method Self-supervised pretraining using contrastive loss with language models.
result Achieved state-of-the-art results on FB15k-237 and FEVER.
Efficiently learns Single-Index Models with constant factor approximation.
problem Learning Single-Index Models under L 2 2 L_2^2 L 2 2 loss with unknown link functions. method An efficient algorithm using alignment sharpness for optimization.
result Achieves constant factor approximation to optimal loss for various distributions and link functions.
Training avoids edge of stability by aligning Jacobian matrices.
problem Training neural networks on the edge of stability causes inaccuracies.
method Used an exponential Euler solver to prevent entering the edge of stability.
result Alignment of Jacobian matrices causes sharpness increase in Hessian.
This paper establishes risk convergence and asymptotic weight matrix alignment --- a form of implicit regularization --- of gradient flow and gradient descent when applied to deep linear networks on linearly separable data. In more detail, for gradient flow applied to strictly decreasing loss functions (with similar re…
Two-layer ReLU networks often converge to simpler solutions, improving generalization.
problem Understanding generalization in overparametrized neural networks, especially for complex tasks.
method Theoretical analysis of two-layer ReLU networks, focusing on the early alignment phase.
result Two-layer ReLU networks often converge to simpler solutions rather than interpolating the training data, leading to better generalization.
Adversarial training makes logistic regression weight loss landscapes sharper.
problem Understanding why adversarial training sharpens the weight loss landscape in logistic regression.
method Theoretical analysis of linear logistic regression model with L2 norm constraints, and experiments on ResNet18.
result Adversarial training sharpens the weight loss landscape in linear logistic regression models.
Algorithm aligns 3D density maps using Wasserstein distance.
problem Aligning 3D density maps in cryogenic electron microscopy.
method Minimizing 1-Wasserstein distance after rigid transformation using Bayesian optimization.
result Improved accuracy and efficiency in protein molecule alignment.
Proposes novel losses for fine-grained categorical domain adaptation.
problem Fine-grained alignment of categories across domains in unsupervised domain adaptation.
method Joint category-domain classifier with adversarial training losses for both domain and category levels, and vicinal domain adaptation.
result Achieves state-of-the-art performance on benchmark datasets.
CUDA CTDR tackles unsupervised domain adaptation without domain alignment.
problem Lack of direct methods for unlabeled target domain classification.
method Jointly learns CTDR on source and target distributions using contradistinguish loss and supervised loss.
result CUDA CTDR achieves state-of-the-art results on various domain adaptation datasets.
Improves domain adaptation by aligning source and target distributions and mitigating noisy labels.
problem Improving performance on target images with different acquisition conditions.
method Combines optimal transport, MixUp regularization, and robust loss for noisy labels.
result Improves domain adaptation performance on various benchmarks and real-world problems.
Hölder-DPO aligns models robustly with noisy human feedback.
problem No existing alignment methods can handle severe label noise.
method Proposes Hölder-DPO, a principled alignment loss with provable redescending property.
result Hölder-DPO enables scalable human feedback valuation and improves model alignment.
Unified deep architecture for domain-invariant network alignment.
problem Eliminate domain representation bias in network alignment.
method DANA (Domain Adversarial Network Alignment) using graph convolutional networks and semi-supervised learning.
result Achieves state-of-the-art alignment results on real-world social networks.
This paper improves model fusion by training-time neuron alignment, reducing barriers in multi-model fusion.
problem Diverse neuron permutations across different settings hinder model fusion performances.
method Training-time neuron alignment using fixed neuron anchors to reduce training-time permutations.
result Training-time neuron alignment improves fusion of pretrained models and federated learning performances.
The paper explains spikes in training loss as catapults, improving feature learning and generalization.
problem Understanding and improving the training process of neural networks.
method Analysis of training loss spikes in SGD, empirical evidence of catapults, and demonstration of AGOP alignment.
result Catapults in training loss promote feature learning and better generalization.
RePULSe improves language model alignment by reducing undesired outputs without sacrificing overall performance.
problem Aligning language models with human preferences while minimizing undesired outputs.
method Integrates probabilistic inference into RL training to reduce undesired outputs.
result RePULSe achieves a better balance between expected reward and undesired output probability.
OOM-RL uses financial market losses to align AI agents in autonomous systems.
problem Constrained alignment of autonomous software agents in live financial markets.
method Deploying agents in live financial markets to enforce strict test-driven workflows.
result Final OOM-RL-aligned system achieved a stable equilibrium with an annualized Sharpe ratio of 2.06.
Paper addresses online alignment of large language models under uncertain preference feedback.
problem Online alignment of large language models with misspecified preference feedback.
method Formulates an oracle-robust objective as a worst-case optimization problem for log-linear policies, and develops projected stochastic composite updates.
result Shows that the robust objective admits an exact closed-form decomposition and achieves O ~ ( ε − 2 ) \widetilde{O}(\varepsilon^{-2}) O ( ε − 2 ) oracle complexity. Local update methods' performance depends on learning rates, affecting convergence rates and alignment with true loss.
problem The performance of local update methods in federated learning and meta-learning is sensitive to learning rates.
method Proved that local update methods perform SGD on a surrogate loss function, characterized the surrogate loss, and derived convergence rates.
result Proper learning rate tuning is crucial for near-optimal behavior in communication-limited settings.
Proposes GDTW for aligning time series on different, incomparable spaces.
problem Dynamic time warping requires comparable spaces, but time series can live on different, incomparable spaces.
method Gromov dynamic time warping (GDTW) considers intra-relational geometry to avoid comparability requirements.
result Demonstrates effectiveness of GDTW in aligning, combining, and comparing time series on incomparable spaces.
Cohen et al. (2021) show GD trajectories align on a bifurcation diagram.
problem Understanding the Edge of Stability (EoS) phenomenon in gradient descent.
method Empirical studies and rigorous mathematical proofs for two-layer networks and single-neuron networks.
result GD trajectories align on a specific bifurcation diagram independent of initialization.
Paper shows equivalence between two alignment methods and introduces a new algorithm.
problem Ensuring human alignment of large language models for useful, safe, and pleasant user experience.
method Introduces IPO-MD algorithm, showing equivalence between IPO and Nash-MD methods.
result Equivalence between IPO and Nash-MD methods proven when considering online version of IPO.
Unified framework simplifies DPO algorithms for LLM alignment.
problem Vast number of DPO variants complicates model alignment.
method Mutual information inspired unifying framework with flexible priors.
result Many DPO variants can be derived from the new framework.
Changing initialization scale affects deep model generalization, leading to memorization or improved performance.
problem Understanding how initialization scale impacts deep model generalization and memorization.
method Experimental setup with varying initialization scales, analysis of activation and loss functions, and development of an alignment measure.
result Increasing initialization scale leads to memorization, and decreasing it improves generalization, depending on activation and loss functions.
A new method for drone-based geo-localization using style and spatial alignment.
problem Geo-localization of drone-view images with satellite-view images using pre-annotated GPS tags.
method Orientation-based method to align patterns, new branch to extract aligned partial features, style alignment strategy.
result The proposed method outperforms state-of-the-art alternatives in geo-localization accuracy.
A new framework improves ASR alignment accuracy via optimal transport.
problem Peaky behavior and alignment inaccuracies in E2E ASR models.
method Differentiable alignment framework based on 1D optimal transport.
result Significant improvement in ASR alignment performance compared to CTC.
A new method reduces preference distortion in LLM alignment.
problem Vulnerability of traditional LLM alignment methods to human preference heterogeneity.
method Sign Estimator: A simple, provably consistent, and efficient estimator using binary classification loss.
result Substantially reduces preference distortion over a panel of simulated personas.
Faster WIND accelerates iterative BOND for LLM alignment.
problem Iterative BOND is inefficient in practice due to sample and computation inefficiency.
method Unified game-theoretic connection to self-play alignment, WIND framework with efficient algorithms.
result WIND variant achieves superior sample efficiency and faster computation.
New method removes pseudo-label bias for unsupervised domain adaptation.
problem Class imbalance and distribution shift between domains.
method Implicit class-conditioned domain alignment without explicit pseudo-label optimization.
result Effective in handling within-domain class imbalance and between-domain class distribution shift.
Recent work on discriminative segmental models has shown that they can achieve competitive speech recognition performance, using features based on deep neural frame classifiers. However, segmental models can be more challenging to train than standard frame-based approaches. While some segmental models have been success…
The paper addresses poor calibration in fine-tuned LLMs after preference alignment.
problem Poor calibration in fine-tuned Large Language Models (LLMs) after preference alignment.
method Proposes a calibration-aware fine-tuning approach to restore calibration without compromising model performance.
result Demonstrates the effectiveness of the proposed methods through extensive experiments.
GENIE balances domain-invariant feature learning and gradient alignment for improved DG performance.
problem Domain Generalization (DG) overfitting to domain-specific features
method GENIE (Generalization-ENhancing Iterative Equalizer) optimizer
result Prevents a small subset of parameters from dominating optimization, promoting domain-invariant feature learning
Enhances hand gesture recognition with separate networks and shared features.
problem Improving recognition accuracy of unimodal 3D-CNNs for dynamic hand gestures.
method Separate networks for each modality, collaborative learning, spatiotemporal semantic alignment loss, focal regularization.
result Improves test time recognition accuracy and state-of-the-art performance.
CoSCA improves unsupervised domain adaptation by better aligning ambiguous target samples.
problem Missing alignment of ambiguous target samples in unsupervised domain adaptation.
method CoSCA explicitly incorporates intra- and inter-class domain discrepancy, estimating label hypotheses and optimizing a contrastive loss with MMD for better global alignment.
result CoSCA outperforms state-of-the-art approaches in producing more discriminative features.
Unified framework connects different neural network models.
problem Understanding the geometry of neural network loss landscapes.
method Unified framework capturing four symmetry classes.
result First discovery of low- and zero-barrier linear interpolation paths.