Search Results - machine+learning

4 Results

Sort By:

Log-Regulated Laplacian Sparse Functional Connectivity of fMRI Data

Description: This technology introduces two novel functional connectivity (FC) estimation methods—a Log-Regularized Laplacian Graph estimator and a K-Nearest Neighbors (KNN) Graph estimator—that generate sparse FC matrices from resting-state fMRI BOLD time series. By pruning weak and irrelevant connections, these methods reduce noise and...

Published: 12/16/2025 | Inventor(s): Mylène Queiroz de Farias, Juliana Danso

Keywords(s): Data, Machine Learning, Neuroimaging

Category(s): Data/AI, Life Science

Uncertainty Estimation for Classification Algorithms through Confusion Matrix Analysis

Description: This technology improves upon traditional point estimates of model accuracy by implementing interval estimation techniques within Frequentist and Bayesian frameworks. Utilizing bootstrap and Monte Carlo sampling methods, it provides a robust measure of uncertainty for classification performance metrics, enhancing the reliability and interpretability...

Published: 12/16/2025 | Inventor(s): Amir Liron, Francis Méndez Mediavilla

Keywords(s): Machine Learning, Software, Statistical Analysis

Category(s): Data/AI, Software

Automated Time Studies using Computer Vision and Large Language Models

Description: This technology uses advanced machine learning algorithms and computer vision techniques to automatically analyze video recordings of work processes, segmenting frames to identify and classify human actions like general movement or tool usage. It employs predetermined time study methods such as Basic MOST to estimate task durations accurately,...

Published: 11/13/2025 | Inventor(s): Abhimanyu Sharotry, Jesus Jimenez, Francis Méndez Mediavilla, Aleah White, Noe Tavira, Lauren Cravy

Keywords(s): Computer Vision, Industrial Process Optimization, Machine Learning

Category(s): Data/AI, Engineering

System and method for identifying maximal independent sets in parallel Photoactivatable Ruthenium Compounds for Cancer Treatment

Description: This technology presents a novel approach to compute maximal independent sets within undirected graphs by leveraging parallel computing resources, such as GPUs. By assigning initial priority values to each vertex based on their degree and the average degree of the graph, the system can simultaneously evaluate and determine the membership...

Published: 12/16/2025 | Inventor(s): Martin Burtscher, Sindhu Devale

Keywords(s): Bioinformatics, Data Mining, Machine Learning, Network Analysis, Network Design, Social Network Analysis

Category(s): Engineering, Data/AI