Esta es nuestra última selección de publicaciones y patentes mundiales en inglés sobre Redes Neuronales, entre muchas revistas científicas online, clasificadas y centradas en red neuronal, neurona artificial, época, arquitectura neuronal, aprendizaje automático, aprendizaje profundo y máquina de vectores de soporte.
Self-supervised optimization and robust back-reconstruction
Published on 2026-06-04 by @MIT
Abstract: AbstractRecent studies have extended nonlinear kernels to Kernel Canonical Correlation Analysis (KCCA), enabling more flexible modeling of complex relationships globally. Building on these developments, we propose three key enhancements to nonlinear KCCA. First, inspired by self-supervised learning in machine learning research, we refine the parameter optimization process by adopting a subject-wise criterion designed to mitigate overfitting. Second, we introduce an improved back-reconstruction ([...]
Our summary: The study enhances nonlinear KCCA by refining parameter optimization with a subject-wise criterion to reduce overfitting. It introduces a robust back-reconstruction method that outperforms existing voxel-importance estimation techniques. The framework is validated through improved performance on simulated and task-based fMRI datasets.
Self-supervised learning, Kernel Canonical Correlation Analysis, back-reconstruction, parameter optimization
Publication
Learning-based segmentation of diffusion-weighted MR images with arbitrary q -space samplings
Published on 2026-06-02 by @MIT
Abstract: AbstractSegmenting anatomical regions is a crucial step in many diffusion-weighted MRI (dMRI) workflows, such as region-of-interest analysis or anatomically-constrained tractography, which enable in vivo studies of brain microstructure and connectivity. However, convolutional neural networks (CNNs)—the foundation of most state-of-the-art segmentation models—require structured inputs with a fixed number of channels. This makes them ill-suited for dMRI, where acquisition protocols vary widely [...]
Our summary: This work presents a novel method for segmenting diffusion-weighted MRI data using geometric deep learning. It directly maps unstructured dMRI data to anatomical segmentations without requiring diffusion model fits. The proposed approach achieves robust generalization and superior performance compared to existing methods.
segmentation, diffusion-weighted MRI, geometric deep learning, convolutional neural networks
Publication
Load monitoring
Patent published on the 2026-05-28 in WO under Ref WO2026109161 by EATON INTELLIGENT POWER LTD [IE] (Reshi Dilpreet Kaur [in], Veluru Suresh [in])
Abstract: A method, apparatus, and computer program are provided for load monitoring for use in an industrial facility comprising a plurality of devices. The method comprises: obtaining energy data associated with the industrial facility, wherein the energy data relates to an aggregate energy consumption of the plurality of devices within the industrial facility over a time period; determining load identification data and initial load disaggregation data for each of the plurality of devices using a respec[...]
Our summary: The method involves obtaining energy data related to an industrial facility s devices. It utilizes a machine learning model to determine load identification and initial load disaggregation data. The final output consists of disaggregated load data for the devices based on the analyzed information.
Load monitoring, energy data, machine learning, load disaggregation
Patent
Artificial intelligence computing device
Patent published on the 2026-05-28 in US under Ref US20260148052 by APACECORE PTE LTD [TW] (Hsu Yi Ting [tw])
Abstract: An artificial intelligence computing device including the following components is provided. A control die is disposed on a substrate. A memory die is positioned above the control die. The memory die includes a dynamic random-access memory (DRAM) for storing a machine learning model. One of the control die and the memory die includes a static random-access memory (SRAM). A deep learning processing unit is electrically connected to the memory die and is configured to execute the machine learning m[...]
Our summary: The device includes a control die and a memory die with DRAM for machine learning. A deep learning processing unit is connected to the memory die. One of the dies also contains SRAM.
AI computing, memory die, deep learning, DRAM
Patent
Test-case generation using a graph model and rag system
Patent published on the 2026-05-28 in US under Ref US20260147693 by INT BUSINESS MACHINES CORPORATION [US] (Wang Wen [cn], Yuan Zhong Fang [cn], Li He [cn], Gao Li Juan [cn], Liu Tong [cn])
Abstract: An example operation includes one or more of extracting testing targets of a software system from a document that describes requirements of the software system, generating a graph model which comprises nodes corresponding to the testing targets and edges between the nodes corresponding to correlations between the testing targets, receiving a request to generate a test case for a testing target among the testing targets, retrieving graph data from the graph model based on the testing target and e[...]
Our summary: This process involves extracting testing targets from software requirements. A graph model is created to represent these targets and their correlations. Machine learning is then applied to generate and execute test cases based on the graph data.
test-case generation, graph model, machine learning, software testing
Patent
Establishing drivers’ trust through event predictability
Patent published on the 2026-05-28 in US under Ref US20260145696 by AUTOBRAINS TECH LTD [IL] (Raichelgauz Igal [il], Cohen Ido [il])
Abstract: [0000] A method for establishing drivers trust through event predictability, the method includes obtaining, at a machine learning process, path information regarding a driving path of a driving by an autonomous vehicle; identifying, by the machine learning process based on the path information, a road scenario that is accommodated, at least in part, in a path segment of the driving path; determining an artificial intelligence model that is below a maturity threshold with respect to providing a d[...]
Our summary: The method focuses on enhancing driver trust through predictable events. It utilizes machine learning to analyze driving paths and road scenarios. The system generates visible indications for drivers when AI decision-making is below a certain maturity level.
machine learning, autonomous vehicles, event predictability, artificial intelligence
Patent
Spatial recall from videos
Patent published on the 2026-05-28 in US under Ref US20260148555 by MICROSOFT TECH LICENSING LLC [US] (Wang Rui [ch], Miksik Ondrej [ch], Yebenes Enric Galceran [ch], Pollefeys Marc Andre Leon [ch])
Abstract: [0000] A technique creates entries in a spatiotemporal data structure that describe objects and activities in videos captured by a plurality of cameras. For instance, each entry in the spatiotemporal data structure includes different kinds of embeddings associated with a particular video captured by a camera. Each entry is further associated with a particular pose in a three-dimensional map and a particular time. In some implementations, the different kinds of embeddings include text embeddings,[...]
Our summary: The technique creates entries in a spatiotemporal data structure to describe objects and activities in videos. Each entry includes various embeddings and is linked to a specific pose and time. A query is processed to retrieve relevant information from the data structure using a neural network.
spatiotemporal data structure, embeddings, neural network, video analysis
Patent
Adaptive modular system and method for ai-driven professional evaluation and benchmarking
Patent published on the 2026-05-28 in US under Ref US20260148174 by NEAL KEVIN [US] (Neal Kevin [us])
Abstract: [0000] This invention relates to an adaptive modular system and method for professional evaluation and benchmarking across multiple industries. The system integrates advanced artificial intelligence and machine learning algorithms within a modular framework to provide personalized, objective, and real-time assessments. Key components include a data ingestion layer for collecting data from diverse sources, a data processing engine for cleaning and transforming data, a feature extraction module, a[...]
Our summary: The invention presents an adaptive modular system for professional evaluation and benchmarking. It utilizes AI and machine learning to deliver real-time, personalized assessments across various industries. Key features include data ingestion, processing, and tailored evaluation modules for enhanced customization and scalability.
Adaptive, Modular, AI, Benchmarking
Patent











