Resumen de: US2024303664A1
In a computer-implemented method of facilitating detection of document-related fraud, fraudulent document detection rules may be generated or updated by training a machine learning program using image data corresponding to physical documents, and fraud determinations corresponding to the documents. The documents and fraudulent document detection rules may correspond to a first type of document. Image data corresponding to an image of one of the physical documents may be received, where the physical document corresponds to the first type of document. By applying the fraudulent document detection rules to the image data, it may be determined that the physical document is, or may be, fraudulent. An indication of whether the physical document is, or may be, fraudulent may be displayed to one or more people via one or more respective computing device user interfaces.
Resumen de: US2024303512A1
In some aspects, the techniques described herein relate to a method including: providing a plurality of inputs to a payload engine; providing a machine learning engine, wherein the machine learning includes a machine learning model, and wherein the machine learning model is configured to generate output based on the plurality of inputs; providing a rules engine, wherein the rules engine includes a logic tree based on the plurality of inputs; receiving, at the payload engine, a transaction and associated transaction details; generating, by the machine learning engine and based on the transaction, the associated transaction details, and the plurality of inputs, a first transaction processing parameter; generating, by the rules engine, and based on the transaction, the associated transaction details, and the plurality of inputs, a second transaction processing parameter; and combining the transaction, the first transaction processing parameter and the second transaction processing parameter into a transaction payload formula.
Resumen de: US2024303384A1
Methods and systems are provided for automating well completion selection and design using machine learning and natural language processing. The present disclosure describes a method for designing a well completion, comprising: i) collecting and storing a historical dataset comprising unstructured data related to prior well completions: ii) identifying a plurality of unstructured schematic documents related to prior well completions that are a part of the historical dataset of i): iii) processing each given unstructured schematic document of the plurality of unstructured schematic documents of ii) to generate structured data corresponding to text of the given unstructured schematic document: iv) associating the structured data corresponding to text of the respective unstructured schematic documents of iii) with different well contexts as part of a database: and v) presenting a graphical user interface to a user for designing a well completion, wherein the graphical user interface presents structured data stored in the database of iv) for insight in designing the well completion.
Resumen de: US2024303553A1
This application describes systems and methods for generating machine learning models (MLMs). An exemplary method includes obtaining a sample and user input data characterizing a product or service. A subset of the data is selected from the sample based on sampling the sample according to the user input data. An MLM is trained by applying the data subset as training input to the MLM, thereby providing a trained MLM to emulate a customer selection process unique to the product or service. A user interface (UI) configured to receive other user input data and cause the trained MLM to execute on the other user input data, thereby testing the trained MLM, is presented. A summary of results from the execution of the trained MLM is generated and presented in the UI. The summary of results indicates a contribution to the trained MLM of each of a plurality of features.
Resumen de: US2024303544A1
A process is provided for using a graph database (e.g., SPOKE) to generate training vectors (SPOKEsigs) and train a machine learning model to classify biological entities. A cohort's input data records (EHRs) are compared to graph database nodes to identify overlapping concepts. Entry nodes (SEPs) associated with these overlapping concepts are used to generate propagated entry vectors (PSEVs) that encode the importance of each database node for a particular cohort, which helps train the model with only relevant information. Further, the propagated entry vectors for a given entity with a known classification can be aggregated to create training vectors. The training vectors are used as inputs to train a machine learning model. Biological entities with an unknown classification can be classified with a trained machine learning model. Entity signature vectors are generated for entities without a classification and input into the trained machine learning model to obtain a classification.
Resumen de: US2024303328A1
Disclosed implementations include a method of detecting attacks on Machine Learning (ML) models by applying the concept of anomaly detection based on the internal state of the model being protected. Instead of looking at the input or output data directly, disclosed implementation look at the internal state of the hidden layers of a neural network of the model after processing of data. By examining how different layers within a neural network model are behaving an inference can be made as to whether the data that produced the observed state is anomalous (and thus possibly part of an attack on the model).
Resumen de: US2024305751A1
Camera platform techniques are described. In an implementation, a plurality of digital images and data describing times, at which, the plurality of digital images are captured is received by a computing device. Objects of clothing are recognized from the digital images by the computing device using object recognition as part of machine learning. A user schedule is also received by the computing device that describes user appointments and times, at which, the appointments are scheduled. A user profile is generated by the computing device by training a model using machine learning based on the recognized objects of clothing, times at which corresponding digital images are captured, and the user schedule. From the user profile, a recommendation is generated by processing a subsequent user schedule using the model as part of machine learning by the computing device.
Resumen de: US2024305689A1
An edge computing platform with machine learning capability is provided between a local network with a plurality of sensors and a remote network. A machine learning model is created and trained in the remote network using aggregated sensor data and deployed to the edge platform. Before being deployed, the model is edge-converted (“edge-ified”) to run optimally with the constrained resources of the edge device and with the same or better level of accuracy. The “edge-ified” model is adapted to operate on continuous streams of sensor data in real-time and produce inferences. The inferences can be used to determine actions to take in the local network without communication to the remote network. A closed-loop arrangement between the edge platform and remote network provides for periodically evaluating and iteratively updating the edge-based model.
Resumen de: US2024303539A1
Embodiments described herein relate to methods and apparatuses for generating one or more answers relating to a machine learning, ML, model. A method in a first node comprises obtaining one or more queries relating to a first output of the ML model, wherein the first output of the machine learning, ML, model is intended to fulfil one or more requirements in an environment; for each of the one or more queries performing a reinforcement learning process. The reinforcement learning process comprises: generating a first set of answers to the query based on the one or more requirements; obtaining a first set of rewards associated with the query, wherein each reward in the first set of rewards is associated with a respective answer in the first set of answers, wherein each reward in the first set of rewards is determined based on one or more metrics; and iteratively generating updated sets of answers associated with the query based on a set of rewards associated with a set of answers from a preceding iteration until a terminal set of answers is reached, in which a second set of answers is generated based on the first set of rewards. Responsive to at least one reward from the reinforcement learning process associated with each query meeting a first predetermined criterion, the method then further comprises initiating implementation of the first output of the ML model in the environment.
Resumen de: WO2024186824A1
In some aspects, the techniques described herein relate to a method including: providing a plurality of inputs to a payload engine; providing a machine learning engine, wherein the machine learning includes a machine learning model, and wherein the machine learning model is configured to generate output based on the plurality of inputs; providing a rules engine, wherein the rules engine includes a logic tree based on the plurality of inputs; receiving, at the payload engine, a transaction and associated transaction details; generating, by the machine learning engine and based on the transaction, the associated transaction details, and the plurality of inputs, a first transaction processing parameter; generating, by the rules engine, and based on the transaction, the associated transaction details, and the plurality of inputs, a second transaction processing parameter; and combining the transaction, the first transaction processing parameter and the second transaction processing parameter into a transaction payload formula.
Resumen de: WO2024184107A1
The disclosure relates to a method for predicting soil characteristic information, the method comprising: receiving data representative of a hyperspectral image of a soil sample; determining a hyperspectral reflectance profile for the soil sample based on the received data, wherein the hyperspectral reflectance profile comprises reflectance intensity information; determining one or more soil characteristics based on the determined hyperspectral reflectance profile of the soil sample, wherein the one or more soil characteristics include one or more of: soil-type classification information for the soil sample; chemical composition information for the soil sample; particle size information for the soil sample A machine learning method for classifying soil samples is also disclosed, along with a training model, and associated systems for carrying out methods according to the disclosure. Unlocking insights from Geo- Data, the present invention further relates to improvements in sustainability and environmental developments: together we create a safe and liveable world.
Resumen de: CN118265985A
A computer-implemented method of managing a first model trained and deployed and used to mark medical data using a first machine learning process. The method includes determining (202) a performance metric for the first model, and if the performance metric is below a threshold performance level, triggering (204) an upgrade process, where the upgrade process includes performing further training on the first model to produce an updated first model, where the first model is configured to update the first model. The further training is performed using an active learning process, in which training data for the further training is selected from a pool of unlabeled data samples according to the active learning process and sent to a reporter to obtain a true value label for use in the further training.
Resumen de: GB2627937A
Methods for retraining only an adaptive portion 330 of a partially-frozen machine learning model 300, such as a neural network or decision tree. In a first method, user equipment apparatus (such as a smartphone) receives a freeze-to-adaptive ratio from network apparatus (such as a server). Based on the freeze-to-adaptive ratio, the user equipment apparatus determines an adaptive portion of the machine learning model to train and a frozen portion 320 of the machine learning model to not train. The user equipment apparatus retrains the adaptive portion of the machine learning model to produce a retrained model. The user equipment apparatus then selects either the retrained model or the original model based on respective performance measures of the two models, such as accuracy or user feedback. In an alternative method, network apparatus accesses a machine learning model comprising a frozen portion and an adaptive portion. The network apparatus generates a backup 340 of the adaptive portion of the machine learning model, retrains the adaptive portion to provide a retrained adaptive portion, and transmits parameters of the retrained adaptive portion to user equipment apparatus. The methods may be used for stateful training of machine learning models in wireless networking.
Resumen de: EP4428770A1
Systems and methods for training machine learning models are disclosed. An example method includes receiving historical event timing data including event data for a first portion including events from a first time period, and a second portion comprising events from a second time period not including the first time period, predicting, based on the first portion of the historical event timing data, a first plurality of predicted events, the first plurality of predicted events corresponding to the second time period, determining a first subset of predicted events to be accurate predictions based at least in part on comparing the first plurality of predicted events to the historical events occurring within the second time period, generating training data based at least in part on the first subset of the first plurality of predicted events, and training the machine learning model based at least in part on the training data.
Resumen de: WO2024181975A1
A method for generating persona specific insights. The method may include receiving sensor data associated with a device; extracting features from the received sensor data; processing the features using a machine learning model to generate machine learning metrics; ingesting the machine learning metrics and the features to generate insights data associated with the device; generating personas data using the insights data and the features, and mapping the insights to the personas data; generating custom insights using the insights data, the personas data, and the features, wherein the custom insights are text-based summaries; and disseminating each of the custom insights to respective persona of the personas data to place service orders associated with the device.
Resumen de: WO2024181895A1
A method performed by a node (100) in a communications network (107). The method comprises: obtaining (202) vulnerability data comprising a plurality of vulnerabilities detected on a host server during a vulnerability scan, matching (204) a first vulnerability in the plurality of vulnerabilities to a first subset of co-occurring vulnerabilities in the plurality of vulnerabilities, the first subset of co-occurring vulnerabilities overlapping in time with the first vulnerability, and training (206) a machine learning model using the first vulnerability as an example input and the first subset of co-occurring vulnerabilities as an example output of the machine learning model.
Resumen de: WO2024182046A1
Certain aspects of the present disclosure provide techniques and apparatus for improved hidden Markov model (HMM)-based machine learning. A sequence of observations is accessed. A hidden Markov model (HMM) comprising a set of transition probabilities, a set of emission probabilities, a transition coefficient hyperparameter, and an emission coefficient hyperparameter is also accessed, and a first output inference from the HMM is generated based on the sequence of observations.
Resumen de: WO2024180015A1
Methods are provided of training a computing module implementing a Machine Learning (ML) model, said methods comprising designating a main computing-node, and performing an iteration loop until an ending condition is satisfied, each of the iterations including: training the computing-module with ML model update using coordinates of a first selection of data-elements in the main computing-node; designating as candidates those computing-nodes parent-child related with the main computing-node based on a candidate-designation policy; calculating, for each of the candidate computing-nodes, a gradient denoting a measure of loss of the ML model using coordinates of a second selection of data-elements in the candidate; and designating as main computing-node one of the candidates depending on the gradients previously calculated, so as to initiate next iteration with said new main computing-node. Training systems and computer programs suitable for performing such training methods are also provided.
Resumen de: WO2024180126A1
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training a predictive machine learning model, wherein the predictive machine learning model is configured to process a model input that comprises an image to generate a predicted image label characterizing the image. In one aspect, a method comprises: obtaining a set of real training examples; generating a set of synthetic training examples for training the predictive machine learning model, comprising: determining an image sampling policy for generating synthetic images based on a distribution of the set of real training examples; generating a plurality of synthetic images, using a generative machine learning model, in accordance with the image sampling policy; and generating a respective synthetic training example based on each of the synthetic images; and training the predictive machine learning model using the set of real training examples and the set of synthetic training examples.
Resumen de: WO2024182155A1
The present disclosure relates to systems and methods for using a director service as an intermediary management system to integrate interactive elements between a developer, a user, a generative machine learning (ML) model, and/or an interactive environment. In examples, the director service may receive input from a user or developer device relating to an interactive element from an interactive environment. The director service may process input from one or more of the developer, the user, and the interactive environment to recognize semantic context and intent objectives associated with the input. The director service may generate one or more prompts based on such input, which is processed by an ML model to generate output. In examples, the prompts may be provided to the ML model to direct it towards providing an output that is responsive to the input and one or more environment guidelines. The input and/or output may be multimodal.
Resumen de: WO2024182178A1
Chemical formulations for chemical products can be represented by digital formulation graphs for use in machine learning models. The digital formulation graphs can be input to graph-based algorithms such as graph neural networks to produce a feature vector, which is a denser description of the chemical product than the digital formulation graph. The feature vector can be input to a supervised machine learning model to predict one or more attribute values of the chemical product that would be produced by the formulation without actually having to go through the production process. The feature vector can be input to an unsupervised machine learning model trained to compare chemical products based on feature vectors of the chemical products. The unsupervised machine learning model can recommend a substitute chemical product based on the comparison.
Resumen de: WO2024182261A1
The present solution provides systems and methods to receive a data set comprising a representation of one or more questions from a survey and provide the data set as input to each of a plurality of machine learning models trained to predict a domain associated with the one or more questions. The systems and methods can receiving as output a first domain prediction for the domain from each of the plurality of machine learning models and determine a second domain prediction for the domain for each question of the one or more questions based on applying a function to each of the first domain predictions. The systems and methods can select, based on the data set and the second domain prediction, an enumerated list of one or more answers from an answer set and cause a display of the enumerated list via a user interface for a selection.
Resumen de: WO2024182095A1
Machine learning based recommendations for user interactions with machine vision systems are provided via populating a graphical user interface (GUI) with a first instance of an image of a product captured by a machine vision system; identifying a feature of the product shown in the image that is associated with a criterion for analyzing the product according to a quality assurance test; identifying, via a machine learning model, a tool for assessing the criterion and settings for the tool based on the feature in the image; populating the GUI with a selectable icon that includes a second instance of the image with an overlay produced according to an assessment of the product via the tool configured according to the settings; and in response to receiving a selection of the selectable icon, adding the tool to a job comprising a series of processes for evaluating the product.
Resumen de: WO2024182276A1
The existential question for Artificial General Intelligence (AGI) is whether the values of AGI will align with human values. Solving "the Alignment Problem" is critical. Get it right, and we unlock trillions of dollars of productivity and huge benefits for humanity. Get it wrong and humanity goes extinct. This invention shows how to design ethical and safe AGI that solves the Alignment Problem. The invention includes scalable ethics and safety features as well as several learning, training, tuning, and customization methods that go beyond the standard techniques of machine learning such as Transformers and Deep Learning techniques. The AGI is implemented using either external problem solvers connected on a network or internal AI agents collaborating within a single computerized system. Detailed implementation examples -- revealing technological and economic synergy with Meta, Amazon, Google, DeepMind, YouTube, TikTok, Microsoft, OpenAI, Twitter/X, Tesla, Nvidia, Tencent, Apple, and Anthropic -- are described.
Nº publicación: WO2024182818A1 06/09/2024
Solicitante:
IQ CONSULTING COMPANY INC [US]
IQ CONSULTING COMPANY INC
Resumen de: WO2024182818A1
Data is the "fuel" that powers the machine learning "engine" for Artificial Intelligence. However, identifying high quality data that can catalyze smarter AI, AGI, and SuperIntelligent systems is becoming an increasingly challenging bottleneck for machine learning. This invention not only describes novel methods for identifying the most valuable data, but it also presents an entirely new framework for understanding the information content of AI-relevant datasets. The methods can be used by intelligent systems autonomously or in collaboration with humans. Novel methods for accelerating AI learning, and for updating the knowledge of AI systems in real-time, are also disclosed. Consistent with the view that human survival may depend on the fastest path to AGI also being the safest path, the invention describes catalysts which help maximize alignment between the values of AGI and humans. These innovative catalysts increase not only the intelligence, but also the safety, of AI systems.