optimal transport
PulseAugur coverage of optimal transport — every cluster mentioning optimal transport across labs, papers, and developer communities, ranked by signal.
11 day(s) with sentiment data
-
New framework optimizes VLM selection and adaptation without target labels
Researchers have developed a new framework called One Stone, Three Birds (OSTB) to address challenges in deploying vision-language models (VLMs) when target annotations are scarce. OSTB uses self-adaptive optimal transp…
-
GFlowNets shown to learn optimal transport plans
Researchers have established a theoretical link between Generative Flow Networks (GFlowNets) and optimal transport (OT). Their work demonstrates that non-acyclic GFlowNets, when optimized, can effectively encode an opti…
-
New geometry framework enables high-dimensional optimal transport
Researchers have developed a new framework called cone-compatible Monge geometry to address challenges in high-dimensional optimal transport. This approach extends the one-dimensional concept of monotone rearrangement t…
-
MESA framework decentralizes LLM safety alignment for MoE models
Researchers have introduced MESA, a new framework designed to enhance the safety alignment of Mixture-of-Experts (MoE) large language models. MESA addresses the issue of "Safety Sparsity" by decentralizing safety respon…
-
New PIBO method optimizes wind farm layouts faster
Researchers have developed a new Bayesian Optimization approach called PIBO, designed to handle problems where the order of elements does not affect the outcome, such as optimizing the layout of offshore wind farms. Thi…
-
New OMT framework offers stable, scalable transport for mixture models
Researchers have developed a new framework called Optimal Mixture Transport (OMT) to address the computational challenges of traditional optimal transport methods. OMT reformulates the problem to operate on mixtures of …
-
New papers explore optimal transport for ML inference
Two new arXiv papers explore advanced inference techniques in machine learning. One paper benchmarks likelihood-free inference methods, evaluating their performance with heavy-tailed and discrete data. The other paper b…
-
New SDR algorithm learns target-aware data representations
Researchers have introduced Supervised Distributional Reduction (SDR), a novel algorithm designed to learn target-aware data representations. SDR combines optimal transport with dependence maximization to balance data c…
-
New framework AvAtar enhances AI alignment with active supervision
Researchers have introduced AvAtar, a new framework designed to improve optimal transport (OT) alignment by actively acquiring supervision. This method quantifies the informativeness of potential supervision points by m…
-
New framework improves bike-sharing demand prediction with temporal adaptation
Researchers have developed a new framework called Gen-ROTDA to improve bike-sharing demand prediction models that degrade over time due to changing travel patterns. This method uses optimal transport to adapt models to …
-
New L2 over Wasserstein framework enhances optimal transport for random measures
Researchers have introduced a new framework called $L^2$ over Wasserstein space to address statistical uncertainty in optimal transport. This framework extends the classical theory to random probability measures, preser…
-
New framework models temporal single-cell RNA data with Gaussian process and optimal transport
Researchers have developed a new generative framework to model temporal processes in single-cell RNA sequencing data. This approach utilizes a latent heteroscedastic Gaussian process, approximated via Hilbert space meth…
-
New geometry and optimal transport methods advance fMRI data analysis
Two new research papers explore advanced geometric and optimal transport methods for analyzing functional magnetic resonance imaging (fMRI) data. The first paper introduces an 'Off-log metric' and Grassmannian subspace …
-
New complexity analysis for normalizing constant estimation in ML
Researchers have developed a new theoretical framework for analyzing the complexity of estimating normalizing constants in probability distributions. This work focuses on annealed importance sampling methods, providing …
-
Transfer learning gains sample efficiency, new paper shows
Researchers have theoretically analyzed the benefits of transfer learning using an optimal transport framework. Their findings suggest that for data dimensions greater than three, transfer learning offers improved sampl…
-
New IC-POT method allows targeted rejection of data in optimal transport
Researchers have introduced a new method called intent-controlled partial optimal transport (IC-POT) to address limitations in existing optimal transport techniques. Unlike traditional methods that enforce exact matchin…
-
New theory analyzes LLM reasoning limits using optimal transport
Researchers have developed a theoretical framework to analyze Large Language Model (LLM) reasoning and out-of-distribution generalization using optimal transport. Their approach quantifies domain shifts with Wasserstein…
-
PACE framework enhances single-cell trajectory inference with geometry-aware transport
Researchers have developed PACE, a new framework for single-cell trajectory inference that addresses the inherent ill-posed nature of reconstructing cellular dynamics from time-course snapshots. PACE utilizes a geometry…
-
New Spherical Transport Method Speeds Up Climate Model Comparisons
Researchers have developed a new method called Spherical Harmonic Optimal Transport (SHOT) to make comparing complex datasets more computationally efficient. This technique adapts optimal transport algorithms for use on…
-
DOODL framework learns shared spectral dynamics across systems
Researchers have developed a new framework called DOODL (Dynamical OperatOr Dictionary Learning) to analyze and learn from multiple related dynamical systems simultaneously. This approach identifies shared structures in…