Stiefel manifold
PulseAugur coverage of Stiefel manifold — every cluster mentioning Stiefel manifold across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
-
New Adam Optimizer Generalizes to Manifolds for Transformer Training
Researchers have developed a novel method to generalize the Adam optimizer to various mathematical manifolds, which are crucial for optimizing neural networks. This new approach, detailed in an arXiv paper, leverages th…
-
New SVGD framework enhances AI model fine-tuning with geometry awareness
Researchers have developed a new framework for parameter-efficient fine-tuning of large pre-trained models that leverages the geometric structure of low-rank manifolds. This approach utilizes Stein Variational Gradient …
-
New online framework for multidimensional functional data analysis unveiled
Researchers have developed a new online framework for functional principal component analysis (FPCA) designed to efficiently model multidimensional functional data streams. This method utilizes tensor product splines an…
-
New COEC framework improves LLM pruning accuracy
Researchers have developed a new training-free framework called COEC (Calibrated Orthogonal-Equivalence Compensation) designed to mitigate accuracy degradation in large language models (LLMs) after structured pruning. C…
-
New research explores diagonal multi-omics integration methods
A new paper published on arXiv introduces methods for integrating heterogeneous datasets, specifically focusing on multi-omics data. The research delves into analyzing biological heterogeneity and develops approaches us…
-
New quantum feedback control method uses physics-informed neural memory
Researchers have developed a new method called Kraus-Parameterized Belief Reinforcement Learning for quantum feedback control. This approach uses a recurrent encoder, constrained to the Stiefel manifold, to generate acc…
-
New arXiv papers explore optimization algorithms for ML
Two new arXiv papers explore advancements in optimization algorithms for machine learning. The first paper introduces Manifold Constrained Steepest Descent (MCSD) and its tangent-projected variant (MCSD-TP) for minimizi…
-
New algorithm enables quantum models to outperform classical HMMs
Researchers have developed a new algorithm called NS-RIS (Newton-Schulz Retraction-based Inference on the Stiefel manifold) for learning Hidden Quantum Markov Models (HQMMs). This method aims to overcome the limitations…
-
New Skewon algorithm offers exact closed-form optimization on Stiefel manifold
Researchers have developed Skewon, a new optimization algorithm for problems involving matrices with orthonormal columns, a common structure in machine learning. This algorithm provides an exact closed-form solution for…
-
New framework enhances uncertainty quantification in reduced-order models
Researchers have developed a new framework for quantifying uncertainty in non-intrusive reduced-order models (NIROMs). This method combines stochastic representation of reduced bases with conformal risk control techniqu…
-
New research explores optimized LoRA fine-tuning methods for LLMs · 4 sources tracked
Researchers are exploring new methods to optimize Low-Rank Adaptation (LoRA) for fine-tuning large language models. One approach, Unified LoRA (ULoRA), introduces a continuum of preconditioned gradient initializations t…
-
New PP-CPCANet framework enhances domain generalization with stable training
Researchers have introduced Projection Pursuit CPCANet (PP-CPCANet), a novel framework designed to improve domain generalization in machine learning. This new method addresses limitations in existing techniques like CPC…
-
New Bayesian Framework Integrates Dimension Reduction for Gaussian Process Models
Researchers have developed a new Bayesian framework designed to address the challenges of Gaussian Process (GP) modeling with high-dimensional inputs. This novel approach integrates dimensionality reduction directly int…
-
ManifoldFlow introduces learnable singular spectrum for neural network weights
Researchers have introduced ManifoldFlow, a novel approach that relaxes the constraints of traditional Stiefel layers in neural networks. This new method allows for learnable singular values, offering greater flexibilit…
-
New method tackles knowledge forgetting in low-rank continual learning
Researchers have identified spectral imbalance as a key factor in knowledge forgetting during low-rank continual adaptation of pre-trained models. They propose a new method that decouples the magnitude of task updates f…
-
Physics-constrained neural networks speed up optical analysis for AR glasses
Researchers have developed a physics-constrained neural network (PCNN) designed for the rapid prediction of outputs from rigorous coupled-wave analysis (RCWA). This novel approach enforces energy conservation as a funda…
-
New method tackles NP-hard diversity selection for large datasets
Researchers have developed a new method called Spectral DPPs via NEPv to address the NP-hard problem of selecting diverse, high-quality subsets from large datasets. This approach recasts the Determinantal MAP objective …
-
New Mirror Descent Framework Extends Optimization to Riemannian Manifolds
Researchers have developed a generalized framework for Mirror Descent (MD) on Riemannian manifolds, extending its applicability to complex optimization problems. This new Riemannian Mirror Descent (RMD) framework includ…
-
New methods tackle black-box optimization with latent space inference and manifold search
Researchers have developed a new method for constrained black-box optimization by reformulating the problem as posterior inference within the latent space of generative models. This approach uses flow-based models and d…
-
Researchers propose novel second-order method for Stiefel manifold optimization
Researchers have developed a novel second-order optimization method for the Stiefel manifold that avoids retractions, offering improved efficiency for high-accuracy requirements. This method combines a tangent component…