35 results about "Self paced" patented technology

The invention provides a segmentation model training method to solve the problems of low training efficiency, long time consumption, and high cost of a segmentation model in the prior art. The method includes: performing free area segmentation on a training sample image by employing an unsupervised free area segmentation method to obtain a free area segmentation image of the training sample image; performing iterative training on a preset convolution neural network by employing a self-paced learning method and by regarding the training sample image as an input image and regarding the free area segmentation image as a mark image to obtain an initial segmentation model; and training the initial segmentation model by regarding a preset standard sample image as the input image and regarding a road area image marked on the standard sample image in advance as the mark image to obtain a target segmentation model. By employing the method, the segmentation accuracy of the segmentation model obtained by training is ensured, the training speed and efficiency can be increased, and road segmentation is performed based on the target segmentation model so that the result of road segmentation is more accurate.

Methods and systems are disclosed for providing learning as a service in a learning network. At least certain embodiments connect users with providers of learning content using an online learning marketplace in a web-based user interface that allows users to control content and pace of learning. The learning can be self-directed, through self-paced online courses, as well as live online courses. The platform is configured to fulfill user requests for learning content without regard to location of the content or time of access. The learning content can be decomposed into monetized components, which can be accessed by users on a per-use or per-session basis.

A computer-operated speech therapy system is provided that includes speech input, speech recognition and natural language understanding, and audio and visual outputs to enable an aphasic patient to conduct self-paced speech therapy autonomously. The system of the invention conducts a therapy exercise by displaying a picture; generating a speech prompt asking the patient for information about the picture; receiving the patient's speech response and processing it to determine its semantic content; determining whether the patient's response was correct; and outputting feedback to the patient. Preferably the system includes a touch screen as a graphical input / output device by which the patient controls the therapy exercise.

It is an object of the present invention to provide a synchronized program delivery of training and self-paced learning a communication network. It is another object of the invention to create a virtual classroom available to clients over a distributed network, thus decreasing the cost of education by eliminating travel costs. The invention provides a delivery environment simulating a classroom including functions for synchronized viewing of multimedia content and shared multimedia objects. A system and method is provided for synchronizing and serving multimedia content in a distributed network. In one aspect there is a synchronization server, a content server, and at least one client. The synchronization server provides indication of an update to a multimedia resource to the client. The content server provides content information to the client based upon the indication provided by the synchronization server.

Systems and methods are provided for enabling a split belt treadmill having two belts to self-pace. A feedback process estimates a first current step speed for a user on each belt by measuring stride length and step duration. A feed-forward process estimates a second current step speed for the user on each belt by measuring three forces and three moment components associated with foot contact with the belt. A command speed for each belt is produced by combining the first and second current step speeds with a Kalman filter. A belt speed associated with each belt is adjusted based upon the command speed.

The invention discloses a semi-supervised pedestrian detection method based on a deep convolutional network, and the method comprises the steps: randomly extracting 5% from a 100% complete training set of a public data set to serve as a training set of semi-supervised learning, reserving the annotations of 5% of the training set, taking the annotations of the 5% of the training set as annotated training sets, taking the rest 95% of the training set as unannotated training sets, and deleting the annotations. An improved RPN convolutional neural network is adopted. The consistency constraint regularization term is added to improve the feature expression capability of the network, so that the robustness of the model is improved; through self-paced learning, the dependency of the deep learningnetwork on a large number of manually labeled samples is reduced, and the over-fitting risk of the deep learning network under the condition of small samples is reduced.

Methods, apparatus, and products are disclosed for self-pacing DMA data transfer operations for nodes in a parallel computer that include: transferring, by an origin DMA on an origin node, a RTS message to a target node, the RTS message specifying an message on the origin node for transfer to the target node; receiving, in an origin injection FIFO for the origin DMA from a target DMA on the target node in response to transferring the RTS message, a target RGET descriptor followed by a DMA transfer operation descriptor, the DMA descriptor for transmitting a message portion to the target node, the target RGET descriptor specifying an origin RGET descriptor on the origin node that specifies an additional DMA descriptor for transmitting an additional message portion to the target node; processing, by the origin DMA, the target RGET descriptor; and processing, by the origin DMA, the DMA transfer operation descriptor.

The invention discloses a self-paced reinforcement image classification method and system. The method comprises the following steps: S10, inputting image data for classification and type labels, and performing feature extraction on the data; S20, based on reinforcement learning and self-paced learning frameworks, establishing a mathematic model; S30, updating parameters of the model and a weak classifier set of the model in an iteration mode until convergence is realized; and S40, predicting types of newly input test images. The method is characterized in that intrinsic consistency and complementarity of a reinforcement learning method and a elf-paced learning method are fully utilized, a distinguishing capability of a classification model and the reliability of image samples participating in learning are simultaneously highlighted in a learning process, and at the same time, effective learning and robust learning are realized. Compared to a conventional image classification method, the method and system have higher classification accuracy and higher robustness for label noise.

The invention discloses a self-paced cross-modal matching method based on subspace. The method is characterized by extracting feature vectors of different modals of data in a data set; obtaining different mapping matrixes corresponding to different modals in a training set through subspace self-paced learning, and mapping the different modal types of data samples in a test set to the same space through the mapping matrixes to enable the data in the training set and the test set to be mapped to a unified space; and then, measuring the similarity between search data and target data in the test set to obtain a cross-modal matching result The method can enable the different modals of data to be mapped to the unified space, and sample selection and feature learning are carried out while mapping, thereby improving matching robustness and accuracy.

An interactive robot-augmented education system designed to promote self-paced, student-driven learning capable of encouraging students using embodied, expressive social interactions, individualized content, attentional tracking and adaptive guidance during each lesson.

The invention discloses a self-paced-cooperative training learning method. The method comprises the following steps of: obtaining data, from two visual fields, of a target problem, and initializing a model, wherein the data comprises a small amount of labeled data and a large amount of unlabeled data; respectively determining corresponding optimization targets on the two visual fields; embedding a self-paced regular term in a loss function of each visual field so as to realize steady learning under the visual field; associating the two visual fields through a regular term; obtaining a multi-visual field semi-supervised self-paced-cooperative training model which is embedded into a steady learning mechanism and has model interpretation; and obtaining high-quality labeling of the unlabeled data by applying the small amount of labeled data and large amount of unlabeled data in a target field and a semi-supervised multi-visual point learning model, and obtaining reliable learning devices under the two visual fields at the same time. The invention aims at providing a steady learning model with a replacement mode for the traditional cooperative training algorithms to ensure that data lack of labeling in the target field can obtain more correct and high-quality labeling.

The invention provides a video object segmentation method based on self-paced weak supervised learning. A self-paced learning algorithm is embedded into a depth neural network, under the guidance of the thought of weak supervised learning, a whole system learns target concepts from the easier to the more advanced, the network obtained by learning with the training process becomes complex, the ability of the network to deal with problems is gradually increased, and finally, an accurate video object segmentation result is obtained. The invention utilizes the advantages of the self-paced learning algorithm and the deep neural network model comprehensively, has higher segmentation accuracy, and shows better robustness when processing video data of different scenes.

The invention discloses a synthetic-aperture-radar (SAR) image change area detection method of self-paced learning. The method mainly solves the problem that in the prior art, sensitivity to speckle noises of synthetic-aperture-radar (SAR) images is prone to cause a loss of part of texture information of the synthetic-aperture-radar (SAR) images, and includes the following specific steps: (1) reading in the synthetic-aperture-radar (SAR) images; (2) carrying out normalization; (3) constructing a change detection matrix; (4) selecting a training sample set; (5) training a deep belief network; (6) constructing a probability matrix; (7) updating the probability matrix; and (8) obtaining a change detection image. According to the method, local information of the original images and self-learning ability of the deep belief network are effectively utilized, the speckle noises are reduced, the local information of the images is retained, and precision of change detection is improved.

The present invention discloses an SAR (Synthetic Aperture Radar) image change detection method based on an SPL (Self-Paced Learning) and a CCN (Convolutional Coupling Network). The problem is solvedthat a current technology is greatly influenced by nose and is not high in detection precision. The method comprises the steps of: (1) reading and processing data; (2) randomly selecting training samples; (3) constructing a convolutional coupling neural network; (4) training the convolutional coupling neural network; (5) employing the self-paced learning method to re-train the convolutional coupling neural network; (6) obtaining feature mapping; (7) employing a logarithm ratio method to obtain a difference chart; and (8) employing a threshold value method to obtain a change detection result map. Compared to the prior art, the SAR image change detection method filters part of training samples with large loss values, inhibits the influence of the noise on the change detection result, reducesthe generation of wrong classification points and improves the classification accuracy.

Disclosed is a method for scoring in real time neural signals of a subject with respect to a reference state characterized by k=1 . . . K reference covariance matrices, the method including the following steps: (i) obtaining neural signals from the subject; (ii) computing a covariance matrix of the neural signals; (iii) computing the Riemannian distances between the covariance matrix and k=1 . . . K reference covariance matrices; (iv) computing a continuous score s in real time based on at least one of the distances computed in step (iii). Also disclosed is a system and method for self-paced modulation or external modulation of neural activity of a subject.

The invention discloses a face age estimation method based on self-paced learning, and belongs to the field of computer vision and machine learning. The method is based on a deep regression forest framework. The method comprises the steps of dividing the face picture into a simple picture (the absolute error between the predicted age and the actual age is small) and a difficult picture (the absolute error between the predicted age and the actual age is large); under a self-stepping learning framework, a strategy from a simple picture to a difficult picture is adopted to train a deep regressionnetwork to establish a nonlinear mapping relation between face features and a target age; finally, performing age estimation on a face image by a random forest, so that the accuracy and robustness ofan existing method are improved. The method can be applied to the aspects of human-computer interaction, age-based safety control, social network entertainment and the like.

The invention discloses a supervised image classification method based on a self-paced constraint mechanism. The method comprises the following steps: dividing difficult types of training samples; establishing a sparse representation model, and substituting samples into the sparse representation model for training; obtaining an image classification model and a prediction model; constructing a category decision-making device, wherein the training sample difficult type comprises a training easy sample and a training difficult sample, and the division training sample difficultly-easy types are divided by adopting a self-step constraint matrix. According to the invention, the training samples are divided through the self-paced constraint matrix; the easy training sample and the difficult training sample are sequentially substituted into a defined sparse representation model for continuous training; a specific self-step constrained image classification scheme can be formed, more judgment information can be conveniently utilized, robustness is achieved on sample noise, and therefore the problem that a supervised dictionary learning mechanism is not suitable for complex samples containingnoise and huge intra-class changes can be solved, and the image recognition effect is improved.

The invention discloses a self-paced learning based infrared spectroscopy quantitative analysis method and system. The method comprises the following steps: step one, extracting an infrared spectroscopy matrix of a standard sample and an infrared spectroscopy matrix of a sample to be detected from the infrared spectroscopy data of the standard sample and the infrared spectroscopy data of the sample to be detected, and obtaining the concentration data of the standard sample; step two, according to the infrared spectroscopy matrix of the standard sample and the concentration data of the standard sample, obtaining the partial least squares weight vector representing the relationship between the e infrared spectroscopy matrix and the concentration data; and step three, according to the partial least squares weight vector, establishing a partial least squares prediction model, and inputting the infrared spectroscopy matrix of the sample to be detected into the partial least squares prediction model to calculate the concentration data of the sample to be detected. Compared with the conventional partial least squares algorithm, the provided method has better stability and generalization ability and a strong anti-noise performance.

Disclosed is self-paced, system-level familiarization training customized to the trainee's individual training needs that is sufficiently flexible to accommodate individuals with varying levels of expertise. Training is accomplished by determining or directing the location of students relative to an on-board system, then stimulating the correct Interactive Media Instruction (IMI) based on that location. Effectiveness is dependent on the factors; location technology, IMI, and movement algorithm; and, the interaction between these three factors.

The invention provides a self-paced matrix decomposition method based on Pareto optimization, and the method comprises the steps: changing a method, which employs a monotone increasing learning step in the conventional self-paced learning to determine the weight of a sample, into a method of obtaining the weight of the sample through the technology of Pareto optimization; changing a sample weight expression range [0, 1] in the Pareto optimization process into a range [-1, 1], and enabling a difficult sample weight to exist in a reasonable distribution range, thereby guaranteeing the diversity; finally selecting a Knee point which achieves the best compromise on an optimization result PF plane as a matrix element weight, thereby speeding up the matrix decomposition process. In a word, the method is more reasonable and accords with the cognitive science in a better way.

Disclosed is an automated tool that assesses an organization's issues through assessments and determines the most appropriate solution flows based on the user's answers and question rank. This invention allows an organization to assess their issues and complete the solution flows in a fully automated environment, providing a less expensive, automated, self-paced flow to improve the organization versus the traditional in person consultation. Any discipline may be incorporated into the tool, allowing new information to be added any time and tailored to specific discipline needs.

The invention provides a pedestrian re-recognition method and system based on self paced feedback. The method comprises the steps that sample areas are divided, and the sample intimacy is calculated;a query sample is input to calculate the score and uncertainty of each sample to acquire a query matching result; at the same time, the quantitative evaluation is carried out on the ranking difficultyof each sample; samples difficult for ranking are selected to form a sample set to be fed back, and a user mark is submitted; after user feedback is completed, a sample mark is updated by the feedback sample set; the ranking score and the difficulty of each sample are recalculated until the final desired search goal of the user is found. According to the feedback technology based on self paced learning and active learning, the information amount of an unlabeled image is fully used; the sample the most difficult for ranking is selected and provided for the user mark; user marks are effectivelyreduced; and the right goal can be accurately and rapidly found.

The invention discloses a self-paced and self-stirred reactor for chloride and concentrated sulfuric acid reaction. Aiming at a process for producing hydrogen chloride gas by adopting chloride and concentrated sulfuric acid as raw materials and feeding the raw materials into a reaction kettle, a self-paced and self-stirred reaction kettle is designed. On one hand, gas generated by reaction is utilized for controlling the feeding speed of the concentrated sulfuric acid through a liquid seal principle, the feeding rate of solid chloride is interconnected controlled according to pressure in the kettle, and the release rate of the reaction heat and a gas flow is eventually controlled. On the other hand, produced cooling gas is utilized for entering from the bottom part of the reactor so as to carry out bubbling stirring, the cooling gas takes away part of reaction heat in the reactor, and reaction foams can be prevented from gathering and blocking a gas outlet, so that the reactor can produce safely, and the energy consumption is reduced. According to the self-paced and self-stirred reactor for the chloride and concentrated sulfuric acid reaction provided by the invention, the feeding speed of the concentrated sulfuric acid is controlled through the liquid seal principle, the feeding rate of the solid salt is interconnected controlled through pressure, and meanwhile, part of the produced gas circularly enters the reaction kettle so as to be bubbled and stirred, so that the equipment is simple, the investment and the energy consumption of the equipment are reduced, and the equipment is convenient for industrial implementation and operation.

A robust multimodal data integration method, termed the SMSPL technique, aimed at simultaneously predicting subtypes of cancers and identifying potentially significant multiomics signatures, is provided. The SMSPL technique leverages linkages among different types of data to interactively recommend high-confidence training samples during classifier training. Particularly, a new soft weighting scheme is adopted to assign weights to training samples of each type, thus more faithfully reflecting latent importance of samples in self-paced learning. The SMSPL technique iterates between calculating the sample weights from training loss values and minimizing weighted training losses for classifier updating, allowing the classifiers to be efficiently trained. In classifying a test sample, outputs of the trained classifiers are integrated to yield a class label by solving an optimization problem for minimizing a sum of classifier losses in selecting a candidate class label, making the SMSPL technique more accruable to discriminate equivocal samples.

The invention discloses a golden monkey face detection method based on increment self-paced learning and regional color quantification. Regional color quantification is used for increasing the difference between a background and monkey body color characteristics so as to more accurately perform segmentation of a monkey body region and reduce the area and the number of the detected suspected monkey body regions. Then face skin suspected region detection is performed in the detected monkey body region. Finally accurate face detection is performed by using increment self-paced learning.

The invention discloses a self-paced learning based infrared spectroscopy quantitative analysis method and system. The method comprises the following steps: step one, extracting an infrared spectroscopy matrix of a standard sample and an infrared spectroscopy matrix of a sample to be detected from the infrared spectroscopy data of the standard sample and the infrared spectroscopy data of the sample to be detected, and obtaining the concentration data of the standard sample; step two, according to the infrared spectroscopy matrix of the standard sample and the concentration data of the standard sample, obtaining the partial least squares weight vector representing the relationship between the e infrared spectroscopy matrix and the concentration data; and step three, according to the partial least squares weight vector, establishing a partial least squares prediction model, and inputting the infrared spectroscopy matrix of the sample to be detected into the partial least squares prediction model to calculate the concentration data of the sample to be detected. Compared with the conventional partial least squares algorithm, the provided method has better stability and generalization ability and a strong anti-noise performance.

Methods, apparatus, and products are disclosed for self-pacing DMA data transfer operations for nodes in a parallel computer that include: transferring, by an origin DMA on an origin node, a RTS message to a target node, the RTS message specifying an message on the origin node for transfer to the target node; receiving, in an origin injection FIFO for the origin DMA from a target DMA on the target node in response to transferring the RTS message, a target RGET descriptor followed by a DMA transfer operation descriptor, the DMA descriptor for transmitting a message portion to the target node, the target RGET descriptor specifying an origin RGET descriptor on the origin node that specifies an additional DMA descriptor for transmitting an additional message portion to the target node; processing, by the origin DMA, the target RGET descriptor; and processing, by the origin DMA, the DMA transfer operation descriptor.