-
公开(公告)号：US20240331684A1
公开(公告)日：2024-10-03
申请号：US18129328
申请日：2023-03-31
CPC分类号： G10L15/16 , G10L15/02 , G10L15/063
摘要： Features of two or more single speaker utterances are concatenated together and corresponding labels of the two or more single speaker utterances are concatenated together. Single speaker acoustic embeddings for each of the single speaker utterances of the concatenated single speaker utterances are generated using a single speaker teacher encoder network. An enhanced model is trained on the concatenated single speaker utterances using a classification loss LCLASS and a representation similarity loss LREP, the representation similarity loss LREP defined to influence an embedding derived from the concatenated single speaker utterances, the influence being based on the single speaker acoustic embeddings derived from the single speaker teacher encoder network.
-
公开(公告)号：US20240331561A1
公开(公告)日：2024-10-03
申请号：US18191908
申请日：2023-03-29
发明人： Martin G. Keen , Jeffrey Bisti , Kriti Kamra , Mark Bylok
IPC分类号： G09B7/02
CPC分类号： G09B7/02
摘要： Disclosed embodiments provide techniques for automated evaluation and scoring of curriculum challenges such as tests and quizzes. An automated score is provided to students, as well as prescriptive guidance on where the provided solution deviates from best practices and/or the taught curriculum. Curriculum material and challenge material are input to a machine learning system to create a grading model. The challenge material can include a software programming challenge. The grading model is used to evaluate curriculum responses and provide a score and feedback based on the evaluation. The evaluation can be based on the curriculum. There can be multiple ways to solve a programming (coding) challenge and disclosed embodiments give scoring preference to solutions that employ techniques covered in the curriculum.
-
公开(公告)号：US20240330983A1
公开(公告)日：2024-10-03
申请号：US18127187
申请日：2023-03-28
IPC分类号： G06Q30/0251
CPC分类号： G06Q30/0261 , G06Q30/0256
摘要： Mechanisms are provided for automatically performing proximity based searches and reminders for items of interest. The mechanisms generate an item listing for a user having entries for things of interest (ToIs) that specify characteristics of the ToIs and notification criteria for notifying the user of availability of the ToIs within a specified proximity of a user location. Location data for locations within the proximity of the user location are processed to identify a provider that matches a characteristic of an entry of the ToI. Provider information is analyzed to determine whether the provider provides the ToI and whether conditions satisfy the notification criteria for the ToI. In response to determining that the provider provides the ToI and the conditions satisfy the notification criteria, an alert notification is output to a user computing device informing the user of the availability of the ToI within the specified proximity of the user location.
-
公开(公告)号：US20240330815A1
公开(公告)日：2024-10-03
申请号：US18127202
申请日：2023-03-28
IPC分类号： G06Q10/0635
CPC分类号： G06Q10/0635
摘要： An embodiment defines a risk model that generates a risk score for an asset based on a plurality of factors, including a first factor that is a first factor type and a second factor that is a second factor type. The model includes a plurality of associations, including a first association that associates a first factor weight with a specified significance of the first factor, and a second association that associates a second factor weight with a time-based metric of the second factor. The embodiment includes modifying one of the plurality of associations resulting in a modified risk model, and generating, using the modified risk model, a risk score for the asset, the generating including determining the risk score for the asset based at least in part on a first factor weight value of the first factor weight and a second factor weight value of the second factor weight.
-
公开(公告)号：US20240330797A1
公开(公告)日：2024-10-03
申请号：US18190179
申请日：2023-03-27
发明人： Sadegh Khalili , Rickie Yeung , Kevan Lucchini , Arkadiy O. Tsfasman , Mary Day
IPC分类号： G06Q10/0631 , G10L15/18
CPC分类号： G06Q10/063114 , G06Q10/063118 , G10L15/18
摘要： A method, computer program product, and computer system are provided for speech recognition-based task organization. Voice data corresponding to a task for a user to perform is received from the user. The task is identified from the received voice data through natural language processing and is classified as a parent task or a child task. The identified task is compared to prior tasks stored within a historical task database. An amount of time needed to complete the identified task is determined based on the compared prior tasks. Actions of the user are tracked and a task progress completion value is determined based on the identified amount of time.
-
86.发明公开公开(公告)号：US20240330715A1
公开(公告)日：2024-10-03
申请号：US18194681
申请日：2023-04-03
发明人： Atul Mene , Jeremy R. Fox , Tushar Agrawal , Sarbajit K. Rakshit
IPC分类号： G06N5/022
CPC分类号： G06N5/022
摘要： A computer-implemented method, a computer program product, and a computer system for amelioration management of edge computing devices. A computer generates digital twin models of respective ones of edge devices. A computer uses the digital twin models to predict an impacted edge device which has a health issue in edge computing. A computer uses the digital twin models to predict impact occurring time. A computer, based on simulations with the digital twin models, reassign a portion of edge computing loads that originally assigned to the impacted edge device to other edge devices. A computer keeps remaining edge computing loads on the impacted edge device such that the impacted edge device is able to complete the remaining edge computing loads prior to the impact occurring time. A computer removes dependencies of the other edge devices on the impacted edge device, in response to the remaining edge computing loads being completed.
-
87.发明公开公开(公告)号：US20240330710A1
公开(公告)日：2024-10-03
申请号：US18194605
申请日：2023-03-31
发明人： Lior Horesh , Cristina Cornelio , Bachir El Khadir , Sanjeeb Dash , Joao P. Goncalves , Kenneth Lee Clarkson
IPC分类号： G06N5/01
CPC分类号： G06N5/013
摘要： A method generates automated discovery of new scientific formulas. The method includes receiving a background theory associated with a phenomenon being studied. The processor receives a set of training data associated with the phenomenon being studied. The set of training data is processed in a machine learning model that generates candidate formulas from data points in the set of training data. Values of a numerical error-vector are generated for the candidate formulas. The candidate formulas are processed in a reasoning model. The operation of the reasoning model includes generating values of a theoretical error-vector based on the background theory. An output of a performance metric is generated based on a generalization of the theoretical error-vector and a reasoning error. The processor determines whether one of the candidate formulas is a meaningful and valid new scientific formula, based on a behavior of the reasoning error and the reasoning performance metric.
-
88.发明公开公开(公告)号：US20240330696A1
公开(公告)日：2024-10-03
申请号：US18192650
申请日：2023-03-30
发明人： Takayuki Osogami , Lan Ngoc Hoang
IPC分类号： G06N3/092 , G06N3/0455
CPC分类号： G06N3/092 , G06N3/0455
摘要： A computer-implemented method for modifying a current policy using reinforcement learning (RL) includes the following operations. A number, corresponding to an inputted sample size, of Markov Decision Processes (MDPs) defining an environment are sampled. For each of the sampled MDPs, behavior data for the current policy is collected, a quantile function of return with the current policy is determined using the collected behavior data, and a current weight is generated by updating a weight for a particular sampled MDP using the quantile function of return for the particular sampled MDP. The policy is modified based upon the weights for each of the sampled MDPs. The current weights are generated by minimizing a conditional value of at risk (CVaR) of a return of the current policy, and the policy is modified to maximize a weighted average of the CVaR of the return with the current weights.
-
公开(公告)号：US20240330554A1
公开(公告)日：2024-10-03
申请号：US18192558
申请日：2023-03-29
IPC分类号： G06F30/333 , G06F30/3323
CPC分类号： G06F30/333 , G06F30/3323
摘要： Scan chain optimization utilizing constrained single linkage clustering is disclosed. In an embodiment, a physical design tool identifies a placement of a plurality of latches in a circuit layout; generates, based on the placement, a set of latch clusters by applying constrained single-linkage agglomerative clustering to the plurality of latches; optimizes the set of latch clusters by redistributing latches across clusters; and generates a set of scan chains corresponding to the optimized set of latch clusters.
-
公开(公告)号：US20240330535A1
公开(公告)日：2024-10-03
申请号：US18190239
申请日：2023-03-27
IPC分类号： G06F30/20
CPC分类号： G06F30/20 , G06F2111/02 , G06F2111/06
摘要： Embodiments of the invention are directed to a programmable computer system having a processor system operable to perform processor system operations that include representing a set of candidate functions in a mathematical expression domain. The set of candidate functions defines relationships between data of an existing system. A set of known background theory is represented in the mathematical expression domain. The set of known background theory defines known relationships associated with the existing system. A model composition operation is performed that includes analyzing, in the mathematical expression domain, the set of candidate functions and the set of known background theory to generate a composed model that satisfies a target data fidelity in a manner that also satisfies a predetermined level of compatibility between the composed model and the set of known background theory.
-
-
-
-
-
-
-
-
-
搜索结果
226673 条记录 (耗时 0.133 秒)
