-
61.发明授权Thompson strategy based online reinforcement learning system for action selection 失效基于Thompson战略的在线强化学习系统的行动选择公开(公告)号：US07707131B2
公开(公告)日：2010-04-27
申请号：US11169503
申请日：2005-06-29
CPC分类号： G06N99/005
摘要： A system and method for online reinforcement learning is provided. In particular, a method for performing the explore-vs.-exploit tradeoff is provided. Although the method is heuristic, it can be applied in a principled manner while simultaneously learning the parameters and/or structure of the model (e.g., Bayesian network model).The system includes a model which receives an input (e.g., from a user) and provides a probability distribution associated with uncertainty regarding parameters of the model to a decision engine. The decision engine can determine whether to exploit the information known to it or to explore to obtain additional information based, at least in part, upon the explore-vs.-exploit tradeoff (e.g., Thompson strategy). A reinforcement learning component can obtain additional information (e.g., feedback from a user) and update parameter(s) and/or the structure of the model. The system can be employed in scenarios in which an influence diagram is used to make repeated decisions and maximization of long-term expected utility is desired.
摘要翻译： 提供了一种在线强化学习的系统和方法。 特别地，提供了用于执行探索与利用的权衡的方法。 尽管该方法是启发式的，但是它可以以原则的方式应用，同时学习模型的参数和/或结构（例如，贝叶斯网络模型）。 该系统包括接收输入（例如，来自用户）并且向决策引擎提供与关于模型的参数的不确定性相关联的概率分布的模型。 决策引擎可以确定是否利用已知的信息，或者至少部分地基于探索与利用权衡（Thompson策略）来探索获取附加信息。 强化学习组件可以获得附加信息（例如，来自用户的反馈）和更新参数和/或模型的结构。 该系统可用于使用影响图进行重复决策的场景，并期望实现长期预期效用的最大化。
-
62.发明授权公开(公告)号：US07689420B2
公开(公告)日：2010-03-30
申请号：US11278899
申请日：2006-04-06
CPC分类号： G10L15/19 , G10L2015/088
摘要： Architecture for integrating and generating back-off grammars (BOG) in a speech recognition application for recognizing out-of-grammar (OOG) utterances and updating the context-free grammars (CFG) with the results. A parsing component identifies keywords and/or slots from user utterances and a grammar generation component adds filler tags before and/or after the keywords and slots to create new grammar rules. The BOG can be generated from these new grammar rules and can be used to process the OOG user utterances. By processing the OOG user utterances through the BOG, the architecture can recognize and perform the intended task on behalf of the user.
摘要翻译： 用于在语音识别应用程序中集成和生成后退语法（BOG）的体系结构，用于识别语法（OOG）语音并更新无上下文语法（CFG）。 解析组件识别来自用户话语的关键字和/或时隙，并且语法生成组件在关键词和时隙之前和之后添加填充标签以创建新的语法规则。 BOG可以从这些新的语法规则生成，并可用于处理OOG用户的话语。 通过通过BOG处理OOG用户话语，架构可以代表用户识别并执行预期的任务。
-
公开(公告)号：US20080154698A1
公开(公告)日：2008-06-26
申请号：US11862683
申请日：2007-09-27
申请人： Gary W. Flake , Lili Cheng , David M. Chickering , Michael Connolly , Alexander G. Gounares , Kamal Jain
发明人： Gary W. Flake , Lili Cheng , David M. Chickering , Michael Connolly , Alexander G. Gounares , Kamal Jain
IPC分类号： G06F17/30
CPC分类号： G06Q30/02 , G06Q30/0201
摘要： The claimed subject matter relates to an architecture that can utilize features of a product to facilitate organization and/or classification of products or product features as well as opinions relating to those products or product features into market identifiers. The market identifiers can aid in aggregating opinions in a more relevant manner that potentially requires less user information about a user in order to achieve bone fide targeting. The architecture can employ data mining techniques to gather information relating to products and opinions thereof in order to create or update data tables and can further allow a user to configure the market identifier in various ways.
摘要翻译： 所要求保护的主题涉及可以利用产品的特征以促进产品或产品特征的组织和/或分类以及与这些产品或产品特征相关的意见进入市场标识符的架构。 市场标识符可以帮助以更相关的方式聚集意见，这可能需要较少的用户信息，以便实现骨骼定位。 该架构可以采用数据挖掘技术来收集与产品及其意见相关的信息，以便创建或更新数据表，并且可以进一步允许用户以各种方式配置市场标识符。
-
公开(公告)号：US08805720B2
公开(公告)日：2014-08-12
申请号：US11959874
申请日：2007-12-19
申请人： Gary W. Flake , Jeffrey R. Hemmen , Alexander Gounares , David M. Chickering , Eric J. Horvitz , Michael Connolly , Lili Cheng , Kamal Jain , George P. Moromisato
发明人： Gary W. Flake , Jeffrey R. Hemmen , Alexander Gounares , David M. Chickering , Eric J. Horvitz , Michael Connolly , Lili Cheng , Kamal Jain , George P. Moromisato
CPC分类号： G06Q30/0224 , G06Q20/209 , G06Q30/02 , G06Q30/0201 , G06Q30/0242 , G06Q30/0255 , G06Q30/0631
摘要： The claimed subject matter relates to an architecture that can facilitate verification or feedback in association with consumer transactions. In particular the architecture can image a transaction receipt, translate the image into text-based data, and extract relevant transaction data there from. Such relevant transaction data can be saved to a data store maintained or supervised by the customer and/or transmitted to authorized components that can benefit from data related to transaction histories, verification, and/or feedback.
摘要翻译： 所要求保护的主题涉及可以促进与消费者交易相关联的验证或反馈的体系结构。 特别地，架构可以对交易收据进行成像，将图像转换为基于文本的数据，并从中提取相关的交易数据。 这样的相关交易数据可以被保存到由客户维护或监督的数据存储器和/或被发送到可以受益于与交易历史，验证和/或反馈相关的数据的授权组件。
-
公开(公告)号：US20110251889A1
公开(公告)日：2011-10-13
申请号：US12757634
申请日：2010-04-09
CPC分类号： G06Q30/0244 , G06Q30/02 , G06Q30/0247
摘要： Various embodiments provide techniques for inventory clustering. In one or more embodiments, a set of inventory to be processed is placed into an initial cluster. The inventory can be related to impressions for advertising that are defined by values for a set of attributes. Recursive division of the initial cluster is performed by selecting an attribute and deriving child clusters that are constrained by one or more values of the attributes in accordance with one or more clustering algorithms. The clustering algorithms are configured to derive an optimum number of clusters by repetitively generating smaller child clusters and measuring a cost associated with adding additional clusters. Additional child clusters can be formed in this manner until the measured cost to add more clusters outweighs a benefit of adding more clusters.
摘要翻译： 各种实施例提供了用于库存聚类的技术。 在一个或多个实施例中，要处理的一组库存被放置到初始集群中。 广告资源可以与由一组属性的值定义的广告展示相关联。 通过选择属性并根据一个或多个聚类算法导出由一个或多个属性值约束的子簇来执行初始簇的递归分割。 聚类算法被配置为通过重复地生成较小的子簇并测量与添加附加簇相关联的成本来导出最佳数量的簇。 可以以这种方式形成额外的子群集，直到添加更多簇的测量成本超过添加更多簇的好处。
-
公开(公告)号：US20110234594A1
公开(公告)日：2011-09-29
申请号：US12748014
申请日：2010-03-26
IPC分类号： G06T11/20
CPC分类号： G06K9/6224 , G06F17/30958 , G06Q30/02 , G06Q30/08
摘要： Various embodiments provide techniques for graph clustering. In one or more embodiments, a participation graph is obtained that represents relationships between entities. An auxiliary graph is constructed based on the participation graph. The auxiliary graph may be constructed such that the auxiliary graph is less dense than the participation graph and is therefore computationally less complex to analyze. Clusters in the auxiliary graph are determined by solving an objective function defined for the auxiliary graph. Clusters determined for the auxiliary graph may then be utilized to ascertain clusters in the participation graph that solve a related objective function defined for the participation graph.
摘要翻译： 各种实施例提供了用于图形聚类的技术。 在一个或多个实施例中，获得表示实体之间的关系的参与图。 基于参与图构建辅助图。 辅助图可以被构造成使得辅助图不如参与图密度小，因此在计算上不太分析复杂。 辅助图中的簇通过求解辅助图定义的目标函数来确定。 然后可以使用为辅助图确定的群集来确定参与图中的聚类，以解决为参与图定义的相关目标函数。
-
公开(公告)号：US07593906B2
公开(公告)日：2009-09-22
申请号：US11461030
申请日：2006-07-31
IPC分类号： G06N7/00
CPC分类号： G06N7/005 , G06Q30/0246 , H04L41/08 , H04L41/147 , H04L41/16
摘要： Enhancements to Bayesian prediction models for network location traffic provide increased accuracy in web traffic predictions. The enhancements include implementing user advertising target queries to determine preferred edges of a Bayesian model, employing hierarchical data structures to cleanse training data for a Bayesian model, and/or augmenting existing data with new training data to enhance a previously constructed Bayesian model. Preferred edge enhancements for the Bayesian model utilize target attribute derived preferred edges and/or explicitly specified preferred edges. Training data is cleansed utilizing tag hierarchies that can employ parent-child relationships, ancestor relationships, and/or network location specific parameters. New training data can also be employed to adjust probabilities in a previously constructed Bayesian model. The new training data can be weighted differently than data represented by the previously constructed Bayesian model.
摘要翻译： 对网络位置流量的贝叶斯预测模型的增强提高了网络流量预测的准确性。 增强包括实现用户广告目标查询以确定贝叶斯模型的优选边缘，采用分层数据结构来清除贝叶斯模型的训练数据，和/或用新的训练数据增强现有数据以增强先前构造的贝叶斯模型。 贝叶斯模型的优选边缘增强使用目标属性导出的优选边缘和/或明确指定的优选边缘。 使用可以使用父子关系，祖先关系和/或网络位置特定参数的标签层次来清理训练数据。 也可以使用新的训练数据来调整先前构造的贝叶斯模型中的概率。 新的训练数据可以与先前构造的贝叶斯模型所代表的数据不同。
-
68.发明申请公开(公告)号：US20090006442A1
公开(公告)日：2009-01-01
申请号：US11769146
申请日：2007-06-27
IPC分类号： G06F7/00
CPC分类号： G06F17/30884 , G06Q10/10 , G06Q30/02
摘要： Improved browsing experience in social bookmarking by leveraging aspects of self tagging and prediction. Quality recommendations are provided for sites of interest to the user and information about what types of people like the current website. Self-tagging is used as an effective means to perform personalized searches. Machine learning and reasoning is employed to predict self-tags based on a website visited and/or website behavior, and self-tags associated with a website and/or webpage based on content of that website and/or webpage. The architecture can be embodied as a browser utility to leverage and extend social-bookmarking information. The utility facilitates the display of information related to a summary view of the users who liked/disliked the current page or website, a tag cloud associated with webpages, and a recommendation button that causes self-tag recommendations to be displayed and that recommends links based on the combination of user tags and content.
摘要翻译： 通过利用自我标记和预测的方面，改善了社会书签的浏览体验。 为用户感兴趣的网站提供质量建议，以及关于什么类型的人像当前网站的信息。 自我标记被用作执行个性化搜索的有效手段。 采用机器学习和推理来基于所访问的网站和/或网站行为来预测自标签，以及基于该网站和/或网页的内容与网站和/或网页相关联的自标签。 该架构可以体现为浏览器实用程序，以利用和扩展社会书签信息。 该实用程序有助于显示与喜欢/不喜欢当前页面或网站的用户的摘要视图相关联的信息，与网页相关联的标签云以及引起自标签建议被显示的推荐按钮，并且推荐基于链接 关于用户标签和内容的组合。
-
公开(公告)号：US20080155576A1
公开(公告)日：2008-06-26
申请号：US11767810
申请日：2007-06-25
IPC分类号： G06F9/44
CPC分类号： G06Q30/0269 , G06F9/451
摘要： The claimed subject matter relates to an architecture or extension to a window manager. In particular, the extension can provide for a window rotation feature that can be exposed as a rotate button. Additionally or alternatively, the rotation feature can be activated based upon a diverse set of conditions, events, and/or commands. Upon activation of the rotation feature, the window manager can rotate a window about an axis to expose a secondary surface that can be populated with alternative content that is distinct from the content of the primary surface of the window. In addition, the architecture provides techniques for identifying both double-sided graphical objects and potentially double-sided objects.
摘要翻译： 所要求保护的主题涉及窗口管理器的架构或扩展。 特别地，扩展可以提供可以作为旋转按钮而被暴露的窗口旋转特征。 另外或替代地，旋转特征可以基于不同的条件，事件和/或命令集来激活。 在激活旋转特征时，窗口管理器可以围绕轴旋转窗口以露出可以用与窗口的主表面的内容不同的替代内容来填充的辅助表面。 此外，该架构还提供了用于识别双面图形对象和潜在双面对象的技术。
-
公开(公告)号：US20080154915A1
公开(公告)日：2008-06-26
申请号：US11769449
申请日：2007-06-27
申请人： Gary W. Flake , Lili Cheng , David M. Chickering , Michael Connolly , Alexander G. Gounares , Jeffrey R. Hemmen , Kamal Jain , Leonard Smith
发明人： Gary W. Flake , Lili Cheng , David M. Chickering , Michael Connolly , Alexander G. Gounares , Jeffrey R. Hemmen , Kamal Jain , Leonard Smith
IPC分类号： G06F17/30
CPC分类号： G06Q30/02 , G06F16/9535
摘要： The claimed subject matter relates to an architecture that can utilize information obtained from a communications system and/or an associated content engine or model in order to facilitate enhanced content recommendations. The information can include content recommendations (e.g., from the content model) as well as information based upon social networking features of the communications system. For example, information such as referrals from friends, family, or other parties that are likely to have firsthand knowledge of interests, objectives, and/or desires of particular consumer that potentially offer a superior data set than conventional data mining by which to form a content recommendation.
摘要翻译： 所要求保护的主题涉及可以利用从通信系统和/或相关联的内容引擎或模型获得的信息的架构，以便促进增强的内容建议。 信息可以包括内容建议（例如，来自内容模型）以及基于通信系统的社交网络特征的信息。 例如，诸如可能具有第一手知识的特定消费者的兴趣，目标和/或欲望的朋友，家庭或其他方的转介的信息，其潜在地提供比常规数据挖掘更好的数据集，以形成 内容推荐。
-
-
-
-
-
-
-
-
-
搜索结果
81 条记录 (耗时 0.04 秒)
