singular value decomposition
PulseAugur coverage of singular value decomposition — every cluster mentioning singular value decomposition across labs, papers, and developer communities, ranked by signal.
3 天有情绪数据
-
LLM analysis method reveals training data secrets and ethical risks
Researchers have developed a method using singular value decomposition (SVD) of a large language model's weight matrix to reveal interpretable semantic subspaces. This technique, requiring minimal code and no model infe…
-
New SVD method simplifies multi-dimensional matching markets
Researchers have developed a novel, computationally efficient mechanism for multi-dimensional matching markets. This new approach uses Singular Value Decomposition (SVD) to simplify complex preference matching into a on…
-
BARRIER framework enables robust machine unlearning via activation geometry
Researchers have introduced BARRIER, a novel framework for machine unlearning that focuses on the geometry of hidden-layer activations rather than static model weights. This approach uses Interval Arithmetic on SVD-base…
-
New frameworks tackle heterogeneous graph learning challenges with decoupled semantics and structure
Researchers have developed a new framework called Decoupled Relation Subspace Alignment (DRSA) to improve the performance of Graph Foundation Models (GFMs) on complex, multi-domain heterogeneous graphs. Existing methods…
-
Federated bandits algorithm slashes computation and communication costs via sketching
Researchers have developed a new method called Federated Sketch Contextual Linear Bandits (FSCLB) to address the computational and communication challenges in federated contextual linear bandits. FSCLB utilizes Singular…
-
New algorithm speeds up EigenDecomposition for large matrices in deep learning
Researchers have developed a new batch-efficient algorithm for EigenDecomposition (ED), a critical computation in computer vision and deep learning. This divide-and-conquer approach aims to overcome the computational bo…
-
Transformer research probes security flaws, training dynamics, and in-context learning limits
Researchers have identified vulnerabilities in the shuffling defense mechanism used to secure Transformer models during inference, demonstrating an attack that can extract model weights by aligning permuted activations.…
-
New method compresses CNNs for medical imaging with improved accuracy
Researchers have developed a novel hierarchical spatio-channel clustering framework to compress convolutional neural networks (CNNs) for medical image analysis. This method partitions feature maps into spatial regions a…