Dies ist unsere neueste Auswahl an weltweiten Veröffentlichungen und Patenten in englischer Sprache zum Thema Prompt Engineering, die in zahlreichen wissenschaftlichen Online-Zeitschriften zu den Themen Prompt Engineering, Prompt Design, Kontextualisierung, Prompt Variabilität, Hyperparameter Tuning, Bias Mitigation, Zero-Shot Learning und Little-Shot Learning veröffentlicht wurden.
Hypernetworks for content recontextualization
Patent published on the 2026-10-15 in US under Ref US20260320860 by EBAY INC [US] (Balicki Jakub Adam [pl], Caudevilla Torras Oriol [us], Drozdz Tomasz [pl], Karanowski Konrad [pl], Lorek Pawel [pl], Lyu Hong [us], Skorupa Aleksander Piotr [pl], Tomczak Jakub Mikolaj [nl], Trzcinski Tomaz [pl], Zieba Maciej [pl])
Abstract: [0000] Some aspects relate to technologies for content recontextualization using hypernetworks to generate Low-Rank Adaptation (LoRA) parameters for adapting a generative model. In some aspects, content recontextualization involves accessing a content item and recontextualization text, generating LoRA parameters by processing the content item with a hypernetwork model, and creating a prompt from the recontextualization text. The generative model, adapted with the LoRA parameters provided by the [...]
Our summary: Hypernetworks are utilized to generate LoRA parameters for adapting generative models. The process involves accessing content and generating prompts from recontextualization text. Training the hypernetwork model requires a dataset of samples for effective parameter updates while keeping the generative model parameters frozen.
Hypernetworks, content recontextualization, Low-Rank Adaptation, generative model
Patent
Contextualization of version control requests utilizing natural language processing for ai based infusion
Patent published on the 2026-09-03 in US under Ref US20260259906 by INT BUSINESS MACHINES CORPORATION [US] (Fox Jeremy R [us], Qiao Mu [us], Keen Martin G [us], Silverstein Zachary Augustus [us])
Abstract: [0000] A system includes a processor that executes computer executable components stored in a memory. The computer executable components can include a tracking component that tracks content changes. The computer executable components can further include a natural language processing component that interprets a natural language query. The computer executable components can further include an artificial intelligence component that contextualizes, based at least in part on the tracked content chang[...]
Our summary: The system utilizes a processor to execute components for tracking content changes. It includes a natural language processing component to interpret queries. An AI component contextualizes these queries based on tracked changes to assess result sufficiency.
version control, natural language processing, artificial intelligence, content tracking
Patent
Combining few-shot learning and meta-adapters for protrusion detection understanding
Patent published on the 2026-08-20 in US under Ref US20260241949 by QUALCOMM INC [US] (Narayanan Venkatraman [us], Ravi Kumar Varun [us], Yogamani Senthil Kumar [us])
Abstract: A method for protrusion detection by a vehicle assistance system includes: obtaining sensor data; generating one or more Bird s Eye View (BEV) features from the sensor data; generating semantic information from the sensor data; applying a first machine learning model to the semantic information to generate refined semantic information that incorporates protrusion information; generating, based on the one or more BEV features and the refined semantic information, a 3D scene representation of a sc[...]
Our summary: This method employs few-shot learning and meta-adapters for effective protrusion detection in vehicle assistance systems. It processes sensor data to create Bird s Eye View features and generates refined semantic information. The system ultimately produces a 3D scene representation to identify detected objects and their protrusions.
few-shot learning, meta-adapters, protrusion detection, machine learning
Patent
Neural temporal fingerprinting and meta-learning for disease detection
Patent published on the 2026-08-20 in US under Ref US20260245724 by BISCHOF KIRSTEN [US] (Chandra Shubham [us], Chandra Gaurav [us], Bischof Kirsten [us])
Abstract: [0000] This invention integrates Neural Temporal Fingerprinting (NTF) with meta-learning to create a dynamic, adaptive system for detecting neurological disorders. By combining CNNs for spectral analysis, GraphSAGE for functional connectivity, and MAML for few-shot learning, the system identifies complex spatiotemporal EEG patterns, enabling accurate detection of Alzheimer s, PTSD, epilepsy, dementia, and comatose states. The system processes real-time or stored EEG data, extracting temporal and[...]
Our summary: This invention combines Neural Temporal Fingerprinting with meta-learning for dynamic disease detection. It utilizes CNNs, GraphSAGE, and MAML to identify EEG patterns for various neurological disorders. The system processes data in real-time, optimizing classification through a reward-penalty mechanism to enhance diagnostics and monitoring.
Neural Temporal Fingerprinting, meta-learning, EEG analysis, neurological disorders
Patent
Machine learning for determining model shard count
Patent published on the 2026-07-30 in US under Ref US20260220525 by INT BUSINESS MACHINES CORPORATION [US] (Dong Lin [cn], Li Qing [cn], Zhao Ze Ming [cn])
Abstract: [0000] A method for processing data in a parallel computing environment comprises: classifying layers of a machine learning model according to layer types; determining a model structure of the machine learning model; setting a shard number parameter as a hyperparameter for each layer of the layers; executing a hyperparameter tuning operation to select a shard number for the shard number parameter according to different model layers and model structures; and applying a loss function to the hyperp[...]
Our summary: This method classifies layers of a machine learning model and determines its structure. It sets a shard number as a hyperparameter for each layer. Hyperparameter tuning is executed to select the optimal shard number based on model layers and structures.
machine learning, model shard count, hyperparameter tuning, parallel computing
Patent
Generative ai industrial automation event based prompt engineering
Patent published on the 2026-07-29 in EP under Ref EP4782989 by ROCKWELL AUTOMATION TECH INC [US] (Briant Stephen C [us], Billi-duran Sharon [us], Miller Michael B [us], Fumagalli Maurizio [us], Kowal Steven J [us])
Abstract: A human-machine interface (HMI) development system leverages a generative AI model to assist in development of HMI projects in accordance with specified functional requirements, which can be provided to the development system as intuitive natural language spoken or written text. The system can formulate and implement HMI project edits during design time based on analysis of this natural language design input. After the HMI project is deployed as a runtime application, the system can also receive[...]
Our summary: A generative AI model assists in developing human-machine interface projects based on functional requirements. The system analyzes natural language inputs to implement design edits during development. It also processes natural language requests for modifications to deployed HMI applications.
Generative AI, Industrial Automation, HMI Development, Natural Language Processing
Patent
Dynamic uncertainty-aware multi-task U-Net for joint lesion segmentation and classification in medical imaging
Published on 2026-07-22 by @OXFORD
Abstract: AbstractSegmentation and classification of lesion is essential for computer-aided diagnosis in grayscale medical imaging, enabling precise localization and reliable categorization of various pathological conditions. However, conventional multi-task learning (MTL) frameworks often suffer from suboptimal task interactions and require manual loss weight tuning, limiting their effectiveness on challenging modalities such as ultrasound, magnetic resonance imaging (MRI), and computed tomography (CT). [...]
Our summary: DynU-Net is a novel multi-task framework for lesion segmentation and classification in medical imaging. It adaptively balances tasks using learnable uncertainty parameters, eliminating manual tuning. Extensive experiments show DynU-Net outperforms existing methods on multiple datasets, achieving high accuracy and efficiency.
Dynamic Uncertainty, Multi-task Learning, Lesion Segmentation, Medical Imaging
Publication
A Method for 3D Building Individualization Integrating SAMPolyBuild and Multiple Spatial-Geometric Features
Published on 2026-02-03 by Lianshuai Cao, Yi Cheng, Zheng Zhang, Ge Zhu, Kunyang Ma, Xinyue Xu @MDPI
Abstract: Individualization of buildings is one of the key issues in the establishment of three-dimensional (3D) building models. Most existing individualization methods rely on inefficient manual separation, while deep learning approaches require extensive pre-training and are highly influenced by the spatial structure of the models. To address these issues, this paper proposes a novel method for 3D building individualization that integrates SAMPolyBuild with multiple spatial-geometric features. Leveragi[...]
Our summary: This paper presents a method for 3D building individualization that integrates SAMPolyBuild with spatial-geometric features. The approach utilizes zero-shot learning for initial extraction and refines accuracy through statistical parameters. Experiments show effective extraction of building models, achieving an F1-score of about 0.83.
3D Building Individualization, SAMPolyBuild, Spatial-Geometric Features, Zero-Shot Learning
Publication
