Uniform Manifold Approximation and Projection
PulseAugur coverage of Uniform Manifold Approximation and Projection — every cluster mentioning Uniform Manifold Approximation and Projection across labs, papers, and developer communities, ranked by signal.
7 day(s) with sentiment data
-
AI models predict welding quality across laser and TIG processes · 5 sources tracked
Researchers have developed advanced deep learning models for predicting weld quality in laser and TIG welding processes. One model utilizes a multi-task spatiotemporal deep neural network to predict penetration depth an…
-
AI framework learns fracture phenotypes without human labels
Researchers have developed a novel label-agnostic framework for characterizing tibial plateau fractures using self-supervised learning. This approach bypasses the need for human-assigned labels, which are prone to inter…
-
AI pipeline automates test spec generation for automotive software
Researchers have developed a novel "Cluster-then-Summarize" pipeline to automate the generation of test specifications for large-scale automotive software requirements. This method embeds requirements, clusters them usi…
-
AI identifies refactoring candidates in BDD test suites
Researchers have developed a novel method to identify and categorize refactoring opportunities within behavior-driven development (BDD) test suites. By employing machine learning classifiers and Large Language Model (LL…
-
LLM latent space geometry visualized using PCA and UMAP
Researchers have developed new methods to visualize the internal geometric structures of large language models (LLMs) by employing dimensionality reduction techniques like PCA and UMAP. Their analysis of GPT-2 and LLaMa…
-
New framework improves trajectory data augmentation for ML
Researchers have developed a systematic framework to improve trajectory data augmentation for machine learning. The study evaluated five selection strategies—Outlierness, Diversity, Representativeness, Uncertainty, and …
-
UMAP embedding improved with new out-of-sample data method
Researchers have developed a new method to improve how new data points are integrated into existing UMAP embeddings. The current UMAP algorithm struggles with out-of-sample points, often placing them incorrectly on the …
-
New IRIS algorithm visualizes time-structured biomedical data
Researchers have developed IRIS, a novel manifold learning algorithm designed to visualize high-dimensional biomedical data that changes over time. Unlike existing methods, IRIS can structure its layouts chronologically…
-
ScaleMAP method preserves density and neighborhood structure in embeddings
Researchers have developed ScaleMAP, a novel method for low-dimensional embeddings that preserves both local density and neighborhood structure. Unlike previous techniques that normalize distances and lose scale informa…
-
Eliot system offers interactive exploration of scientific literature trends
A new system called Eliot has been developed to help researchers navigate the rapidly expanding volume of scientific literature. Eliot interactively explores trends by retrieving arXiv papers in real-time, clustering th…
-
MEDAL framework enables quantitative validation of manifold embeddings
Researchers have introduced MEDAL (Manifold Embedding Distillation via Autoencoder Learning), a new framework designed to quantitatively validate manifold embeddings. MEDAL distills existing embeddings into an encoder-d…
-
New method tackles ambiguous data points in dimensionality reduction
Researchers have introduced a novel graph-based method to address ambiguous instances in dimensionality reduction, a common source of visual artifacts. This approach identifies data points that are highly similar to mul…
-
ML classifier automates refactoring of BDD test suites
Researchers have developed a method to automatically identify and categorize opportunities for refactoring in behavior-driven development (BDD) software test suites. Their approach uses machine learning classifiers, spe…
-
New framework detects manipulative political narratives on social media
Researchers have developed a new framework to detect and categorize manipulative political narratives found on social media. The system first uses a few-shot prompt with a reasoning model to filter out manipulative post…
-
LLMs help score and cluster urban bridge importance using graph analysis
Researchers have developed a new method to assess the importance of urban bridges using heterogeneous graph analysis and large language models. This approach quantifies bridge importance based on factors like transit ac…
-
UMAP dimensionality reduction method compared to PCA and t-SNE
A new paper compares Uniform Manifold Approximation and Projection (UMAP) with other dimensionality reduction techniques like PCA and t-SNE. The study systematically evaluates supervised UMAP for both regression and cla…
-
New Class Angular Distortion Index metric improves dimensionality reduction faithfulness
Researchers have introduced the Class Angular Distortion Index (CADI), a novel metric for evaluating dimensionality reduction techniques. CADI addresses limitations in existing metrics by assessing the faithfulness of c…
-
VERA tool automatically explains 2D data embeddings with region annotations
Researchers have developed VERA, a new method for automatically generating visual explanations of two-dimensional data embeddings. VERA identifies key regions within these embeddings and links them to human-interpretabl…
-
Diffusion Transformer generates synthetic fraud data to improve detection
Researchers have developed a new diffusion model called EmDT, designed to generate synthetic data for fraud detection. This model utilizes UMAP clustering to identify specific fraud patterns and a Transformer network to…
-
Topology tool Mapper reveals how language models encode ambiguity
Researchers have introduced Mapper, a topological data analysis tool, to better understand how language models handle ambiguity. Applied to RoBERTa-Large, Mapper revealed that fine-tuning reorganizes the model's embeddi…