150 results about "Knowledge transfer" patented technology

A computer-implemented method executed by at least one processor for training fast models for real-time object detection with knowledge transfer is presented. The method includes employing a Faster Region-based Convolutional Neural Network (R-CNN) as an objection detection framework for performing the real-time object detection, inputting a plurality of images into the Faster R-CNN, and training the Faster R-CNN by learning a student model from a teacher model by employing a weighted cross-entropy loss layer for classification accounting for an imbalance between background classes and object classes, employing a boundary loss layer to enable transfer of knowledge of bounding box regression from the teacher model to the student model, and employing a confidence-weighted binary activation loss layer to train intermediate layers of the student model to achieve similar distribution of neurons as achieved by the teacher model.

A learning environment and method which is a first milestone to an expert machine that implements the master-slave robotic concept. The present invention is of a learning environment and method for teaching the master expert machine by a skilled worker that transfers his professional knowledge to the master expert machine in the form of elementary motions and subdivided tasks. The present invention further provides a stand alone learning environment, where a human wearing one or two innovative gloves equipped with 3D feeling sensors transfers a task performing knowledge to a robot in a different learning process than the Master-Slave learning concept. The 3D force\torque, displacement, velocity\acceleration and joint forces are recorded during the knowledge transfer in the learning environment by a computerized processing unit that prepares the acquired data for mathematical transformations for transmitting commands to the motors of a robot. The objective of the new robotic learning method is a learning process that will pave the way to a robot with a “human-like” tactile sensitivity, to be applied to material handling, or man / machine interaction.

The invention discloses a neural network training method and device so as to solve a technical problem that a student network trained by the existing knowledge transfer mode cannot give consideration to a large application range and accuracy. The neural network training method comprises the steps of selecting a teacher network which realizes the same functions as the student network; and iteratively training the student network based on distribution of first intermediate layer features and second intermediate layer features corresponding to matching with the same training sample data so as to acquire a target network, and thus transferring knowledge of intermediate layer features of the teacher network to the student network, wherein the first intermediate layer features are a feature map outputted from a first specific network layer of the teacher network after inputting the training sample into the teacher network, and the second intermediate layer features are a feature map outputted from a second specific network of the student network after inputting the training sample data into the student network. The neural network acquired by adopting the technical scheme disclosed by the invention is not only wide in application range but also excellent in performance.

A method for efficiently determining and displaying pertinent information determined from multiple input and calculated parameters associated with a videoconference call. The method for efficiently determining and displaying this personal information is performed using input from the user at an endpoint and calculated information throughout the videoconference to present personal information, about the currently speaking person, to all participants. Videoconferencing systems are typically used by multiple people at multiple locations. The method of this disclosure allows for more user interaction and knowledge transfer amongst the participants. By sharing information between the different locations participants are more aware of who is speaking at any given time and the importance to be applied to what that particular person is saying.

The invention discloses a small sample depth learning method based on knowledge transfer of a shallow model. The invention firstly preprocesses the data, and then transforms the original signal into different transform domains according to the prior knowledge and expert experience of the related field, and calculates the artificial features. According to artificial features, different shallow models are selected and trained based on a small amount of labeled sample data. According to classification accuracy / prediction error and other indicators, different shallow models with different featurecombinations are screened to form candidate model pool. Then, based on the candidate model pool, the model is selected to predict the unlabeled samples, and the prediction tags are obtained, and a plurality of prediction tags are fused. The prediction tags are combined with a small number of existing tagged samples to construct the training set. Deep neural network structure is designed for the specific task, and the training is based on the above-mentioned mixed training set. The validity of the proposed method is verified by the rotating machinery fault diagnosis data set.

A shareable utility that provides quality information to users in a given subject area. The utility maintains a quality database of information on the subject area on a local computing device. The quality database has links to objects of interest residing on the Internet and on the local device. All of the information in the quality database has been recommended for inclusion by an expert in the field. The present utility provides for dual searching of the local computing device and the Internet. A primary screen with multiple secondary windows is used as the starting point. The primary screen is designed to provide structure and direction to the search, but may also be customized to include a user / predefined template representing a broad level of interest. Subsequent display screens are used to display more detailed levels of interest in the given subject area. The present utility is well suited for applications in education, and knowledge transfer applications (e.g. specific code routines, best practices, lessons learned, etc.). The present utility architecture is highly adaptable, reusable and supports searches of all kinds (e.g. financial statements 2004, branch info, etc.). By aggregating expert knowledge (of “where” to look), other vendors search capabilities and the digitized resources available, one can leverage this utility to maximum advantage.

The invention provides an image text description method based on a knowledge transfer multi-modal recurrent neural network. According to the method, by a knowledge transfer model in a multi-modal unit, identification capacity of an existing image classifier on most of objects and grammatical structures and semantic association in an existing corpus are well utilized, a target object in an image can be more accurately described, and generated sentence descriptions can be richer in grammatical structure, appropriate in semantics, and higher in readability.

A system for knowledge transfer in a group setting comprises a plurality of participant work areas and a moderator work area. Each of the plurality of participant work areas has at least one corresponding participant input-device, and each of the participant input-devices is adapted to define participant images that are then included on the corresponding participant work area. The moderator work area comprises at least one moderator input-device. The at least one moderator input-device is adapted to define moderator images that are then included on the moderator work area and to select moderator images that are then simultaneously included on each of plurality of participant work areas. The moderator input-device is further adapted to select participant images from any of the plurality of participant work areas that are then included on the moderator work area.

The invention provides a cross-domain and cross-category news commentary emotion prediction method. According to the method disclosed by the invention, under the condition that a target domain is provided with a small amount of annotation data only and another related but different source domain is provided with a large amount of annotation data, knowledge transfer among different domains is realized through simulating the relationship between the emotion category collections of the source domain and the target domain, and a cross-domain and cross-category news commentary emotion prediction model is built, so that the problem of difficulty in emotion prediction of news commentaries of the target domain is solved; under the situation that the emotion category collections of the source domain and the target domain are different, the method disclosed by the invention is significantly better than other alternative cross-domain and cross-category online news commentary emotion prediction methods, and high cost resulting from manual annotation work and energy consumed through training more classification models are greatly reduced. The method can be applied to user sentiment analysis and public sentiment supervision.

A zero-sample classifying method based on class transfer comprises the steps of acquiring a vision characteristic of C kinds of training samples, a class semantic characteristic of the training sampleand a true label matrix; calculating a semantic similarity matrix by means of cosine similarity or Gaussian similarity through the class semantic characteristic; calculating a diagonal matrix of a class semantic similarity matrix; calling a Sylvester equation in an MATLAB toolset for obtaining a mapping matrix; inputting the vision characteristic of the training sample, the corresponding class semantic characteristic and the true label matrix into a target function, continuously adjusting the value of a model regularization parameter, calculating the least value of the target function, and finishing model training; and in a testing period, inputting the vision characteristic of the testing sample and the corresponding semantic characteristic, calculating scores of the classes, and determining the class with highest score as the predicated class of the testing sample. The zero-sample classifying method based on class transfer has advantages of sufficiently digging the semantic relationbetween different classes, realizing knowledge transfer between a known class classifier and an unknown class classifier, and realizing high convenience in application in image classification.

The disclosure relates to transferred learning from a first language (e.g., a source language for which a semantic parser has been defined) to a second language (e.g., a target language for which a semantic parser has not been defined). A system may use knowledge from a trained model in one language to model another language. For example, the system may transfer knowledge of a semantic parser from a first (e.g., source) language to a second (e.g., target) language. Such transfer of knowledge may occur and be useful when the first language has sufficient training data but the second language has insufficient training data. The foregoing transfer of knowledge may extend the semantic parser for multiple languages (e.g., the first language and the second language).

The invention provides a 5G Internet of Things intrusion detection method and system based on federated transfer learning. The 5G Internet of Things intrusion detection method comprises the steps of:collecting a to-be-detected traffic in the Internet of Things, and acquiring a feature vector of an original data packet; and inputting the feature vector into a corresponding client model fm, k, andjudging whether the to-be-detected traffic is legal or not. According to the 5G Internet of Things intrusion detection, the federation transfer learning method in the 5G Internet of Things IDS is proposed for the first time, data from different Internet of Things can be safely aggregated, a good intrusion detection model for each Internet of Things is realized through knowledge transfer and sharing, the 5G Internet of Things intrusion detection method can be conveniently and safely applied to various different Internet of Things, and the method has very strong generalization ability and is suitable for popularization and application. Compared with an existing method, abnormal traffic can be detected more accurately, and unknown attacks can be detected more effectively.

The invention relates to a knowledge acquiring system field, in particular to a knowledge intelligent acquiring system of vegetable supply chains, aiming at designing a set of knowledge intelligent acquiring system of vegetable supply chains, realizing the effect knowledge spreading of the vegetable supply chains by means of the computer media, and particularly realizing the knowledge transfer in the field of the vegetable supply chains at home and aboard. The system takes the ontology model of the vegetable supply chains as a support and provides three retrieval modes, i.e. keyword retrieval, semantic extension retrieval and semantic retrieval based on the ontology model as a tool for acquiring the knowledge in the professional field. The system has high query efficiency and facilitates the participants of the vegetable supply chains and scientific researchers to acquire the professional knowledge.

Provided are a speech recognition training processing method and an apparatus including the same. The speech recognition training processing method includes acquiring a multi-talker mixed speech signal from a plurality of speakers, performing permutation invariant training (PIT) model training on the multi-talker mixed speech signal based on knowledge from a single-talker speech recognition model and updating a multi-talker speech recognition model based on a result of the PIT model training.

A computer implemented system and method for providing a computer and collaboration platform around knowledge transfer, expertise, innovation, tangible, intangible and information assets are disclosed. The system converts a static expert content into an active forum in an effective manner to promote collaboration among users in the various categories. The system converts the static content into one or more sections according to a parsing rule. Each section is assigned with one or more categories and one or more plug-ins for forming a framework. The system captures the framework as a model and generates one or more data partnering objects, and stores them in a database. Pursuant to a user's request and attributes of the user, appropriate one or more of the data partnering objects are retrieved from the database to generate an active forum. The system, then, publishes the active forum.

An educational maze navigated through responses to questions provided at points in the maze enhances factual assimilation and retention for the user. The user selects responses to travel through the maze and is not immediately informed whether the selected responses are correct. Various indications can be provided to inform the user about the correctness of responses or whether they are on a path towards the maze exit. The indications may be dead ends, loops, suggested reference materials or other information that tends to assist the user in assimilating facts or navigating the maze. The user may receive a reward for providing correct responses or successfully completing the maze. The problem-solving and cognitive judgment skills applied to navigate the maze tends to help users assimilate and retain the factual information represented by the queries and responses.

Methods and systems for facilitating online knowledge transfers. The methods and systems may include receiving content, including learning opportunities and learning information from a plurality of sources and transforming the source format of the received content into a common system format when the source format of the content does not match the system format. In addition, the methods and systems include filtering and indexing the content and populating a data repository with the indexed content in the system format. The methods and systems include transmitting a content feed with the content in the system format for generation on an output mechanism.

The invention provides a bridge bearing disease identification method based on transfer learning between convolutional neural networks. The method comprises the steps that bridge bearing disease photos are acquired, and label information is assigned to each photo; an image processing method is used to increase the volume of data used for training the convolutional neural networks; all pictures ina training set and a test set are scaled to color pictures of a predetermined size, and image preprocessing is carried out; a convolutional neural network model which is trained on other data sets isacquired; and a knowledge transfer method is used to acquire a convolutional neural network model with the function of automatic bridge bearing disease identification. According to the automatic bridge bearing disease identification method based on the convolutional neural networks, the convolutional neural networks trained in a knowledge transfer mode have obvious advantages in accuracy and convergence speed; the volume of data required for training the neural networks is greatly reduced; and the method is of practical significance for bridge bearing diseases with complex disease scenes and difficult data collection.

This invention discloses an excavation method of generalized association rule used to analysis the compound rule of Chinese medicine; the method is essentially an excavation of association rule combined with excavation method of generalized association rule expressed by field knowledge, using semantic Web technology as the principal means expressed by field knowledge. The method relates to a knowledge discovery device, the device use nomenclature system and field rule provided by the knowledge database to complete the data excavation process and submit the excavation results to knowledge database in the form of knowledge proposal, then certified and evaluated by the field experts. The data excavation process comprises: extract required data from Chinese medicine data resource first, then excavate senseful frequency model in the data and make semantic labels, extract and reason the generalized rules according to the labeled model. The use of semantic Web technology fields to construct knowledgebase, and the attainment of information and knowledge transfer between the components the methods referred to.

The invention discloses a migration sparse coding image classification method based on dictionary domain adaptation, belonging to the machine learning field. This method introduces dictionary alignment mechanism based on traditional sparse coding model and constructs feature transfer classification model to solve the problem of image cross-domain classification. When the samples in source domain and target domain obey different distributions, the traditional sparse-coded image classification algorithm can not learn from the samples in source domain to obtain the dictionary which can encode thesamples in target domain effectively, so the classification performance is degraded. At first, that dictionary alignment mechanism is introduce into a sparse coding model, then the constraint term ofthe dictionary are converted into an unconstrained optimization problem by L2 regularization, and the knowledge transfer performance of the model is improved by adopting an inter-domain dictionary approximation as a regularization term. The invention can effectively extract cross-domain image sparse feature representation and obtain higher classification accuracy.

A method for transferring knowledge including the steps of organizing an author's source documents including presentation and reference material into units of presentation and units of reference, establishing one or more sets of rules of presentation of the units, defining one or more components of presentation, preparing one or more sequences of presentation of the units, storing the units of presentation, units of reference, sequences of presentation, component definitions, and sets of rules in a knowledge database on a source computing system, and generating, based on the sets of rules, a collection of interlinked browser-ready electronic document pages.

The invention discloses a human body behavior identification method based on thematic knowledge transfer. The method comprises the steps that a bilingual dictionary at a training visual angle and a test visual angle is built, wherein the bilingual dictionary is used for transforming low-layer features of the same action at the two visual angles to the same representation; three steps of low-layer feature extraction, middle-layer feature extraction and bilingual dictionary obtaining are included; all action videos at the training angle are adopted, lower-layer features of different actions at the training angle are transformed to representations respectively through the bilingual dictionary, and classified models recognizing the different actions are trained; test action videos at the test visual angle are adopted, lower-layer features of actions at the test angle are transformed to representations through the bilingual dictionary, and reorganization results of the actions are obtained through the classified models. The human body behavior identification method based on thematic knowledge transfer significantly improves the recognition rate of human body behaviors at the crossed visual angles, has high robustness for change of the visual angles, and has significant value in video monitoring.

A migration system and method. The migration method comprises receiving a request to generate a migration plan for an entity and generating a strategic plan comprising a first procedure for transferring knowledge required for performing job functions comprised by a first set of individuals to a second set of individuals and a second procedure for defining a set of temporary individuals to perform a transition management process through the transferring. Additionally, an allocation plan and a ramping down plan are generated. The allocation plan is for allocating the set of temporary resources to the entity for the transition management process and second allocating said second set of individuals for the transferring. The ramping down plan is for ramping down to close out the set of temporary individuals based on external data. The resource migration plan is for supporting the entity and includes a procedure for allocating the second set of individuals for supporting the entity.

The invention relates to the field of combination of deep learning and machine vision in artificial intelligence, in particular to a knowledge transfer learning-based aerial power part image classification method, which comprises the following steps of establishing a convolutional neural network GoogLeNet; adjusting and optimizing the convolutional neural network GoogLeNet, on the basis of the convolutional neural network GoogLeNet, replacing the last three layers of the convolutional neural network GoogLeNet with a full connection layer, a softmax layer and a classification output layer, andcarrying out optimization setting; when the network is trained, obtaining the network parameters by combining multiple simulation experiments and a Bayesian optimization algorithm; performing normalization preprocessing on the acquired electric power part images, inputting the images into the set new deep convolutional neural network obtained in the step 2 for learning, and performing classification according to the types of insulators, hardware fittings, towers and the like; and verifying by performing simulation experiments.

The invention relates to a cross-media retrieval method based on a hybrid migration network. The method comprises the following steps that 1, a single-media database and a cross-media database are set up, and data in the cross-media database is divided into a training set and a test set; 2, the hybrid migration network is trained by means of data in the single-media database and the training set of the cross-media database, and used for learning unified representation of different media data; 3, by means of the trained hybrid migration network, unified representation of the data in the test set of the cross-media database is obtained, and then the cross-media similarity is calculated; 4, one media type in the cross-media test set is used as a query set, the other media type serves as a retrieval database to conduct retrieval, and the final retrieval result is obtained according to the similarity. Accordingly, knowledge transfer from single-media to cross-media is achieved, the unified representation more suitable for cross-media retrieval is generated by emphasizing semantic association of a target domain, and the accuracy rate of the cross-media retrieval is increased.

Embodiments of the present disclosure provide a method and apparatus for document recommendation by obtaining a plurality of first data for a source recommendation task from activities related to a source user using a plurality of first documents; obtaining a plurality of second data for a target recommendation task from activities related to a target user using a plurality of second documents; performing the target recommendation task based on the plurality of first data, the plurality of second data, and knowledge transferred from the source recommendation task to obtain a target recommendation model; and conducting document recommendation to the target user using the target recommendation model.

A knowledge management system to optimize the human learning and memorization process. Memorization of desired information is achieved through knowledge transfer and rehearsal, whereby a learning system provides a series of associated elements to the user using a dynamical and reactive algorithm that responds to the user's abilities to prepare an efficient rehearsal process. The instant system combines a dynamic learning system with a database that comprises content provided by multiple users and which is overseen by individuals to ensure the validity of the content within the system. Information may be provided in a multimedia format which can include music, video, text and other auditory and visual stimuli as part of the rehearsal process.

An embodiment of digital watermarks which contains literary information of interest to a reader of an electronic literary macramé (ELM) or a knowledge transfer tool (KTT). The nature of the watermark changes by the deepening of its role in the ELM / KTT, since the watermark becomes what is termed in cryptographic parlance a “side channel” for conveying narrative-related information to a reader outside the primary communications stream. If a reader removes the digital watermark from the work, the side channel containing narrative-related information is destroyed, and the literary value of the work is thereby diminished.

A computer implemented system and method for providing a computer and collaboration platform around knowledge transfer, expertise, innovation, tangible, intangible and information assets are disclosed. The system converts a static expert content into an active forum in an effective manner to promote collaboration among users in the various categories. The system converts the static content into one or more sections according to a parsing rule. Each section is assigned with one or more categories and one or more plug-ins for forming a framework. The system captures the framework as a model and generates one or more data partnering objects, and stores them in a database. Pursuant to a user's request and attributes of the user, appropriate one or more of the data partnering objects are retrieved from the database to generate an active forum. The system, then, publishes the active forum.