96 results about "Context modelling" patented technology

Techniques are disclosed for using a combination of explicit and implicit user context modeling techniques to identify and provide appropriate computer actions based on a current context, and to continuously improve the providing of such computer actions. The appropriate computer actions include presentation of appropriate content and functionality. Feedback paths can be used to assist automated machine learning in detecting patterns and generating inferred rules, and improvements from the generated rules can be implemented with or without direct user control. The techniques can be used to enhance software and device functionality, including self-customizing of a model of the user's current context or situation, customizing received themes, predicting appropriate content for presentation or retrieval, self-customizing of software user interfaces, simplifying repetitive tasks or situations, and mentoring of the user to promote desired change.

An entropy encoding / decoding scheme uses an arithmetic encoder / decoder together with a context modeler for performing a context-based adaptive binary arithmetic coding, which is especially suited for video compression applications. The context modeler includes an initializer for initializing the context modeler by determining and providing initialization probability information to the arithmetic encoder, the initialization probability information to be used by the arithmetic encoder for processing a start symbol of a symbol sequence. The initializer is operative to determine the initialization probability information based on an estimation of the symbol statistics relating to the start symbol such that an initialization probability distribution is different from an equi-probable distribution for all symbols of the symbol set. By means of the initialization, the context-based adaptive binary arithmetic coding / decoding scheme is optimally adjusted before processing a start symbol of the symbol sequence so that no bit rate compression losses because of a delayed context-based adaptation are introduced.

An entropy encoding / decoding scheme uses an arithmetic encoder / decoder together with a context modeler for performing a context-based adaptive binary arithmetic coding, which is especially suited for video compression applications. The context modeler includes an initializer for initializing the context modeler by determining and providing initialization probability information to the arithmetic encoder, the initialization probability information to be used by the arithmetic encoder for processing a start symbol of a symbol sequence. The initializer is operative to determine the initialization probability information based on an estimation of the symbol statistics relating to the start symbol such that an initialization probability distribution is different from an equi-probable distribution for all symbols of the symbol set. By means of the initialization, the context-based adaptive binary arithmetic coding / decoding scheme is optimally adjusted before processing a start symbol of the symbol sequence so that no bit rate compression losses because of a delayed context-based adaptation are introduced.

Corpus analysis methods have previously been applied to text, typically to annotated text. The invention shows how to apply corpus analysis methods to information captured in databases, where the database columns include a mixture of both structured domains and unstructured domains containing text. It uses case-based methods to automatically organize cases for periodic review. The invention can help to identify opportunities for increasing knowledge about databases. By organizing a database around common lexical, semantic, pragmatic and syntactic relationships, the invention can be used to increase the effectiveness of previous corpus analysis methods, and to apply them to a diversity of commercial applications. The invention applies contextual constraints to focus the application of linguistic methods. This invention can provide a component for medical records, enterprise databases, information retrieval, question answering systems, interactive robots, interactive appliances, linguistically competent speech recognition, speech understanding and many other useful devices and applications that require a high level of linguistic competence within operational contexts.

A novel method for going beyond the observed properties of a keyword, to a model in which the presence of a term in a document is assessed not by looking at the actual occurrence of that term, but by a set of non-independent supporting terms, defining the context. In other words, similarity is determined not by properties of the keyword, but by properties of the keyword's context. This yields a scoring for documents which is useful for ad hoc retrieval and, by extension, any information retrieval task where keyword-based similarity is needed.

A method for context modeling of coding information of a video signal for compressing or decompressing the coding information is provided. An initial value of a function for probability coding of coding information of a video signal of an enhanced layer is determined based on coding information of a corresponding video signal of a base layer.

A method for context-modeling coding information of a video signal for compressing or decompressing the coding information is provided. An initial value of a function for probability coding of coding information of a video signal of an enhanced layer is determined based on coding information of a video signal of a base layer.

A method for context-modeling coding information of a video signal for compressing or decompressing the coding information is provided. An initial value of a function for probability coding of coding information of a video signal of an enhanced layer is determined based on coding information of a video signal of a base layer.

A method for context-modeling coding information of a video signal for compressing or decompressing the coding information is provided. An initial value of a function for probability coding of coding information of a video signal of an enhanced layer is determined based on coding information of a video signal of a base layer.

This invention provides a novel single-pass and multi-pass synchronized encoder and decoder, performing order(s) of magnitude faster data compression and decompression, at any compression ratio with the higher or the same perceived and measured decompressed image quality in comparison with the best state-of-the-art compression methods, using order(s) of magnitude less system resources (processor complexity, memory size, consumed power, bus bandwith, data latency). These features are achieved using novel direct and inverse non-stationary filters for the recusive octave direct and inverse subband transformation, novel simple context modeling and symbol probability estimation using a minimum number of histograms with the fast adaptation for the sign and the magnitude of the transformation coefficients, a novel accelerated range coder without division operations, and a novel synchronisation of the compressed data.

A method for context-modeling coding information of a video signal for compressing or decompressing the coding information is provided. An initial value of a function for probability coding of coding information of a video signal of an enhanced layer is determined based on coding information of a video signal of a base layer.

A method for context-modeling coding information of a video signal for compressing or decompressing the coding information is provided. An initial value of a function for probability coding of coding information of a video signal of an enhanced layer is determined based on coding information of a video signal of a base layer.

The invention discloses a parallel non-zero coefficient context modeling method for binary arithmetic coding, and relates to a context modeling technology for video coding. The method is proposed to solve the problem that the data throughput rate of a coding system is reduced because the conventional binary arithmetic coding generates a data dependency relationship on the context in the context modeling process of non-zero coefficients. The method comprises the following steps of: 1, defining the number of coefficients and non-zero coefficients in a transform quantification block; 2, performing binarization on the non-zero coefficients to obtain a bin sequence; 3, performing context modeling on a first context according to the position information of the non-zero coefficients and the number of the non-zero coefficients in the transform quantification block; 4, calculating the probability distribution of the non-zero coefficients, the absolute value of which is abs (Li) in the first context value; 5, subtracting 1 from the absolute value of Li, and performing binarization; and 6, performing context modeling by using the equal probability distribution. By using the method, the context modeling processes of different non-zero coefficients can be simultaneously performed, and parallel execution of multiple context modeling processes in the coding process is realized.

According to the invention, a data model and method and apparatus for performing content and context modeling are disclosed. The method dynamically classifies and gathers selective information on various monitored systems to detect content related problems and provide context for diagnosing the root cause of these problems. The selected, monitored information for classification is converted to a plurality of dimensions that may be preconfigured, added incrementally after the monitored system is in production, or when a need for more advanced analysis or for wider context arise.

In various embodiments of the present invention, a context-based denoiser is applied to each noisy-image symbol embedded within a context to determine a replacement symbol for the noisy-signal symbol. The context-based denoiser includes a context-modeling component that efficiently generates context classes and symbol-prediction classes, assigns individual contexts to context classes and symbol-prediction classes, collects symbol-occurrence statistics related to the generated context classes and symbol-prediction classes, and, optionally, generates noisy-symbol predictions.

A novel method for going beyond the observed properties of a keyword, to a model in which the presence of a term in a document is assessed not by looking at the actual occurrence of that term, but by a set of non-independent supporting terms, defining the context. In other words, similarity is determined not by properties of the keyword, but by properties of the keyword's context. This yields a scoring for documents which is useful for ad hoc retrieval and, by extension, any information retrieval task where keyword-based similarity is needed.

A method and apparatus of video and image coding using context modeling for video and image are disclosed. A selected context is determined from a context set for a first syntax element at a first location based on statistics of one or more second syntax elements at one or more second locations within a predefined region. The first syntax element is different from the second syntax element. The first syntax element is then encoded or decoded according to the selected context. In one embodiment, the context modelling is applied to transform coefficients. The first syntax element corresponds to coeff_abs_level_greater1_flag and the second syntax element corresponds to significant_coeff_flag.

Systems and methods are provided for improving language models for speech recognition by adapting knowledge sources utilized by the language models to session contexts. A knowledge source, such as a knowledge graph, is used to capture and model dynamic session context based on user interaction information from usage history, such as session logs, that is mapped to the knowledge source. From sequences of user interactions, higher level intent sequences may be determined and used to form models that anticipate similar intents but with different arguments including arguments that do not necessarily appear in the usage history. In this way, the session context models may be used to determine likely next interactions or "turns" from a user, given a previous turn or turns. Language models corresponding to the likely next turns are then interpolated and provided to improve recognition accuracy of the next turn received from the user.

A context modeler models a context representation and a method models a context representation. The context modeler models a context representation in an application. The context is represented in a current situation by at least one context entity that is included in at least one collection of references to a plurality of entities in a database in accordance with the situation. The context modeler includes an activation model for assigning an activation attribute to the at least one context entity indicating the importance of the at least one context entity in the current situation. The context modeler takes into account the activation attribute in modeling the context representation.

The invention discloses a pre-training text abstract generation method based on neural topic memory, and the method comprises the steps: carrying out the coding of a sequence according to a complete input sequence through employing the context modeling capability of BERT pre-training, and completing the text embedding; encoding potential topic representations by using a topic memory network according to the output sequence representations; performing matching according to theme representation, reasoning and the coded sequence representation to form a final coded representation, and then usinga decoder for generating a preliminary output sequence; and masking each word in the output sequence and feeding into the BERT, and in combination with the input sequence, a decoder based on transformer and LSTM is used to predict the refined word of each mask position to realize fine adjustment. According to the method, the BERT layer and the LSTM layer of the bidirectional context are used for deeply capturing the features, the masked abstract is fed, the parameters are finely adjusted, and each abstract word is refined, so that the text abstract which is smoother and high in information amount is generated.

A method of signaling of coding modes including an IntraBC mode (Intra-block copy mode) to improve the coding efficiency is disclosed. In one embodiment, the partition mode associated with IntraBC mode and the corresponding partition mode for Inter mode use the same binarization process and the same context modelling. In another embodiment, the displacement vector of IntraBC prediction for the chroma block is derived from the displacement vectors of the corresponding luma block. In yet another embodiment, a non-square PU is partitioned into multiple TUs (transform units) and IntraBC prediction is performed TU by TU.

This invention provides a novel single-pass and multi-pass synchronized encoder and decoder, performing order(s) of magnitude faster data compression and decompression, at any compression ratio with the higher or the same perceived and measured decompressed image quality in comparison with the best state-of-the-art compression methods, using order(s) of magnitude less system resources (processor complexity, memory size, consumed power, bus bandwith, data latency). These features are achieved using novel direct and inverse non-stationary filters for the recusive octave direct and inverse subband transformation, novel simple context modeling and symbol probability estimation using a minimum number of histograms with the fast adaptation for the sign and the magnitude of the transformation coefficients, a novel accelerated range coder without division operations, and a novel synchronisation of the compressed data.

A method and apparatus for applying entropy coding to a symbol associated with a block are disclosed. According to the present invention, context-based entropy coding is applied to source symbols associated with blocks having variable block sizes generated by partitioning an initial block using a quadtree structure, a binary-tree structure or a combined quadtree plus binary-tree structure. Contexts according to the present invention are based on some information derived from neighbouring blocks and also based on at least one of the shape, the size and the depth of the current block since the statistics of the symbols associated with the current block may be correlated with how the current block has been partitioned through a tree structure. A current symbol to be encoded or decoded may correspond to split flags and modes associated with the tree structure, skip flag or prediction mode flag.

The invention relates to a method for apperceiving context based on a mobile agency, which utilizes mobile agency technology to apperceive the context; context information is obtained, modeled, stored and utilized through the mobile agency; the method has applicability, intelligence and reusability; the method comprises the following specific steps: a, when a system is initialized, a mobile agency platform dispatches an executive body agency to an intelligent executive body layer and dispatches an evolution agency to a context modeling layer respectively; b, when a user inquires the context information or customizes a system, the mobile agency platform dispatches an inquiry agency or a customization agency to a context obtaining layer; c, the inquiry agency or the customization agency accesses corresponding data source according to the type of the context information; and a relational database is accessed through a JDBC interface; d, the evolution agency is combined with a reasoning engine, uses ontology reasoning to remove inconsistency of an original context and is mapped to an ontology library; and e, the executive body agency processes the received data and selects a proper mode to provide service for the user through self-adaptation, self-organization and other strategies.

An entropy decoding module includes a binary arithmetic coding module that generates a bin string by pipeline processing a bit stream, based on a clock signal and based on context model information. A binarization and context modeling module generates a stream of syntax elements and the context model information based on the bin string.

The present invention provides a method for context modeling of coding information of a video signal for compressing or decompressing the coding information is provided. An initial value of a function for probability coding of coding information of a video signal of an enhanced layer is determined based on coding information of a corresponding video signal of a base layer.

In one embodiment, a method for encoding video data is provided that includes receiving a transform unit comprising a two-dimensional array of transform coefficients and processing the transform coefficients of the two-dimensional array along a single-level scan order. The processing includes selecting, for each non-zero transform coefficient along the single-level scan order, one or more context models for encoding an absolute level of the non-zero transform coefficient, where the selecting is based on one or more transform coefficients previously encoded along the single-level scan order.

An image compression apparatus includes an image transformer for transforming each pixel of an image data into binary bit array. The bits of a same bit plane are located in a same column. A context modeler, coupled to the image transformer, includes a context model template and a context model. When the bit array is imported, a plurality of reference context bits for a bit to be compressed is determined by means of the context model template. The bit to be compressed and the plurality of reference context bits are on a same bit plane. A mathematic encoder, coupled to the context modeler, encodes the bit to be compressed according to the context modeler and the information provided by the plurality of reference context bits.

The invention discloses a multi-model integrated target detection method with rich space information. The multi-model integrated target detection method combines a single-stage framework, context modeling and multi-scale representation, and is applied to a network model for target detection in an integrated way. Specifically, a new context modeling method is adopted, for applying hole convolutioncommonly used in the semantic segmentation field to target detection, and a context detection module is constructed by utilizing the characteristic that the hole convolution can expand a receiving field under the condition that the calculation amount is not increased; meanwhile, fine-grained details are captured through multi-scale representation, so that the representation capability of the modelis enhanced; and the thought of ensemble learning is combined, so that the performance of the detector is further improved.