52 results about "Predictive learning" patented technology

A method comprises receiving a network of a plurality of nodes and a plurality of edges, each of the nodes comprising members representative of at least one subset of training data points, each of the edges connecting nodes that share at least one data point, grouping the data points into a plurality of groups, each data point being a member of at least one group, creating a first transformation data set, the first transformation data set including the training data set as well as a plurality of feature subsets associated with at least one group, values of a particular data point for a particular feature subset for a particular group being based on values of the particular data point if the particular data point is a member of the particular group, and applying a machine learning model to the first transformation data set to generate a prediction model.

An adaptive computer-implemented e-learning system and method able to deliver a learning approach that customizes and adapts to each learners' skills and knowhow. A computer-implemented cognitive diagnosis process for predictions on skills and knowledge of a learner comprising the steps of: building a domain ontology; establishing a items bank; referencing the items bank using elements from the domain ontology elements (semantic-grounding of the items); establishing a training database of learners answers; estimating Cognitive Diagnostic Model (CDM) parameters and determining an appropriate model to be used; running the model to estimate a learner level on non-abstract latent variables; applying a Mastery of domain Knowledge and competency Skill (MKS) approach to estimate the learner level on abstract latent variables, or extracting relevant hierarchies from the ontology and use the CDM appropriate model to estimate those attributes; and building a diagnosis report from the results of steps 6) and 7).

The invention discloses an intelligent dynamic route inducible system. The system comprises a traffic flow process, a detection unit, an identification unit, a predictive learning unit, a prediction unit, an inductive learning unit, a decision-making unit, an evaluation unit and a knowledge base. Compared with the prior art, the system provides more better inductive strategies, and realizes intelligent route induction of a traffic road network more efficiently, under the conditions that besides providing traffic induction strategies according to the traffic flow process detected in real time, a traffic induction system can complete knowledge reservation of inductive strategies in an intelligent learning mode for the inductive strategies superior to standard state in the system through a learning process, and the inductive strategies obtained through learning can be combined to the traffic flow detected in real time.

The invention provides a practical effect evaluation and learning path recommendation method based on cognitive diagnosis. The practical effect evaluation and learning path recommendation method comprises an expansion feature preprocessing model, a deep knowledge tracking model, a knowledge network construction model and a path recommendation algorithm based on cognitive ability. The expansion feature preprocessing model performs primary evaluation of cognitive competence according to skill attributes in a learner test process, and introduces personalized difference information into the diagnosis model. And the deep knowledge tracking model predicts the knowledge mastery ability of the learner according to the test sequence and the heterogeneous features of the implicit coding, and the knowledge mastery ability serves as a basis of learning guidance. The exercises and knowledge network construction model provides a global map of scientific thinking, and the cognitive diagnosis is combined to recommend a learning path to a learner, so that the cognitive ability difference in the learning process is considered, and the logic of a knowledge structure is followed.

A microprocessor controlled automated, multi-fuel apparatus to blend hydrogen, bio-fuel and / or natural or propane gases. This novel multi-stage apparatus first converts cooking oils into bio-fuel. The system automatically blends the bio-fuel with at least one of or both hydrogen gas, generated by a self-contained on-board hydrogen electrolyzer, and / or natural or propane gases. This blended “Hyenrich” gaseous fuel drives various processes including, but not limited to, cogeneration systems and electrical generators to produce “green” electricity by utilizing an adaptive and predictive learning algorithms to significantly reduce cost per kilowatt and lessen dependency on the over taxed utility grid, while simultaneously reducing emissions of CO, CO2 and NOx, making the method and apparatus an environmentally-friendly energy device.

The invention belongs to the technical field of intelligent education, and discloses a personalized test paper composition method and system fusing cognitive characteristics and test question text information, and the method comprises the following steps: firstly predicting the score of a learner on a specific test question based on the cognitive level through a cognitive diagnosis model; predicting scores of the learners on the specific test questions based on text information by using a recurrent neural network model; constructing a probability matrix decomposition target function on the basis of the obtained learner based on the cognitive level and the prediction score of the text information, and predicting the potential score of the learner on the specific test question; and finally,calculating KL divergence by utilizing the estimated learner knowledge mastering vector and the learner incremental knowledge mastering vector, and selecting test questions with increased learner knowledge mastering trend and proper difficulty to form personalized test paper in combination with the potential score of the learner on the test questions. According to the invention, the test paper forming result can be customized according to the test target and the test question difficulty, and the autonomous learning efficiency of learners is greatly improved.

The invention provides an artificial intelligence method and system for calculating geological parameters by utilizing logging-while-drilling data. Big data pre-processing is carried out through a software-defined artificial system; data learning of predictive learning and ensemble learning is included; indicative learning of data action guidance is realized based on a Merton's law; by means of amulti-branch synchronous learning system and method, exploration on logging data over time is solved by utilization of machine predictive learning; exploration on the logging data in terms of spatialdistribution is solved by utilization of ensemble learning; and the generation direction of geological parameters to be calculated is solved by utilization of indicative learning. According to the artificial intelligence method and system provided by the invention, geological parameter calculation is carried out through downhole logging data; therefore, stratigraphic structure parameters and electrical parameters capable of describing the geological occurrence are obtained; the data transmission amount in a drilling process can be greatly reduced; simultaneously, geological information reflected by the logging data is visual and quantitative; and thus, the artificial intelligence method and system have the important significance for geological orientation and logging interpretation.

The invention provides an intelligent teaching system based on deep learning, and the system comprises a learning record obtaining module which is used for obtaining learning records from a database, and the related information comprises student numbers, test question numbers, related knowledge points, answering results, and answering time; and a scene coding module which is used for summarizing, summarizing and coding past learning records and current scenes of the students to obtain general codes. A learning sequence coding module which is used for coding previous learning sequences of the knowledge points by the students to obtain knowledge point answering codes; a neural network training module which is used for training a neural network according to the general bracket codes and the knowledge point answer codes; a prediction module which is used for predicting the answering result of the student in a given scene; and a learning scheduling module which is used for recommending learning contents to the students according to prediction of the answering results of the students.

The invention discloses an identification-increasing-degree super-regression load-modeling multi-curve fitting model based on a support vector machine. The model includes a single training set fittingfunction creating module, a multi-increment learning set fitting function creating module, an extrapolation prediction learning set creating module, a minimum vector spacing optimizing module and a multi-curve fitting function creating module; the single training set fitting function creating module is oriented to a single curve observation value training set and based on an algorithm of a support vector regression machine and obtains fitting functions of non-linear objects; the multi-increment learning set fitting function creating module is oriented to multiple curve observation value training sets and obtains multiple corresponding fitting functions; the extrapolation prediction learning set creating module is used for creating sets of all output vectors as identification increasing degree learning sets; the minimum vector spacing optimizing module adopts the minimum vector spacing as an optimizing index for searching for an aggregation center of the identification increasing degree learning sets on the basis of an identification increasing degree learning set, and the center serves as a data training set representing comprehensive characters of all curves; the multi-curve fitting function creating module obtains fitting functions involving the basic features of all the curves. The model based on the support vector machine can achieve the purpose of making local feature identification degree gradually approach to whole feature identification degree in a fitting process.

The invention belongs to the technical field of education big data mining, and discloses a multi-modal unified intelligent learning diagnosis modeling method and system, a medium and a terminal. The method comprises the steps: constructing a multi-channel cognitive diagnosis model, performing preliminary diagnosis on learners, and performing parameter estimation on learning resources to obtain a learning resource parameter set and a learner parameter set; performing modeling on the learning resources and learners to obtain depth representation features; introducing a self-attention mechanism to fuse learner features and learning resource features; taking the fusion features as a data basis for predicting the performance condition of the learner, and constructing a learner performance prediction network to obtain a predicted value of the correct answer probability of the learner; and diagnosing the overall knowledge point mastering condition of the learner according to the characteristic information of the learner and the exercises, and acquiring parameter characterization of the exercises. The advantages of the multi-channel cognitive diagnosis model can be fused, the neural network is designed to carry out intelligent learning diagnosis on the learner, and expandability is achieved.

The invention discloses a method for predicting the residual life of a rotating machinery under multiple working conditions based on a dynamic domain adaptation network. The method comprises the following steps: 1, generating a source domain sample set and a target domain sample set; 2, preprocessing vibration signals in the source domain sample set and the target domain sample set; 3, generating a target domain training set and a target domain test set; 4, selecting a source domain training set by adopting a reverse verification technology; 5, constructing a dynamic domain adaptive neural network which structurally comprises a feature extractor, a prediction learning module, a marginal distribution adaptive module and a conditional distribution adaptive module; 6, training the dynamic domain adaptive neural network to obtain a trained dynamic domain adaptive neural network model; and 7, predicting the residual life of a target domain test set by using the model. According to the method, the generalization ability and the prediction precision of the residual life prediction model are improved under the condition of multiple working conditions.

The invention discloses an online education-oriented learner abnormal learning state prediction method. The method comprises the steps of preprocessing high-dimensional online education platform log information and learner registration information, and coding and constructing learner portrait features based on a self-supervised learning method; constructing state features of the learner, further constructing a state feature sequence based on a generation time sequence of the state features, and constructing a state feature graph based on cosine similarity among the state features; constructing a long-short term memory-graph attention deep network conforming to learning badness degree prediction of online education, and determining the number of layers of the network, the number of neurons of each layer and the dimensions of input and output; constructing a pseudo tag based on the noise tag to perform iterative training on the network; and predicting the abnormal learning state and degree of the learner in the to-be-predicted learning stage by using the trained network. According to the method, the state abnormity degree of the learner is predicted by using the learner registration information and the learner log information, and a reference is provided for a teacher to carry out targeted guidance and help on the learner.

The invention provides a service quality evaluation system based on sub-time window deep reinforcement learning. The system comprises a data acquisition module, a model adjustment module, a reward feedback module, a parallel learning module, a Q table updating module, a periodic iteration module, a prediction learning module and a time window adjustment module. According to the method, the problems of low evaluation accuracy, poor timeliness and the like in an existing quality evaluation method are solved by utilizing the interactivity and decision-making ability of reinforcement learning andthe perception ability of deep learning.

The invention discloses a knowledge tracking method and system fusing question difficulty. The method comprises the following steps: acquiring a question qt and an answer rt answered by a learner at a t moment and interaction information when the learner answers the question; obtaining a concept matrix Mk and a concept mastering matrix at the moment t, and calculating question difficulty dt according to the interaction information; obtaining the relevancy wt (i) of the question qt and the ith concept; calculating to obtain the overall mastering degree kst of the learner to the concepts related to the question qt; outputting the prediction accuracy prt of the question qt answered by the learner and the prediction difficulty pdt of the question qt for the learner; training a prediction model, wherein the training target is to minimize the difference between the prediction accuracy prt and the answer rt and the difference between the prediction difficulty pdt and the question difficulty dt; and updating the concept mastering matrix of the next time step. While predicting the accuracy of the learner, the invention also predicts the difficulty of the question to the learner, and enhances the expression of the model in the aspect of the knowledge mastering state of the learner.

The invention discloses a network equipment fault prediction method, which comprises the following steps of: obtaining a system alarm log, and extracting a parameter characteristic value from the alarm log; marking the time slice as a positive sample and a negative sample according to whether a fault occurs based on the parameter characteristic value; sending the positive sample and the negative sample into an online random forest, and respectively training the positive sample and the negative sample based on the positive sample hyper-parameter and the negative sample hyper-parameter to obtaina prediction model; and acquiring a real-time log, extracting a parameter characteristic value from the real-time log, and inputting the parameter characteristic value into the prediction model for prediction. The invention further discloses a corresponding device, computer equipment and a readable storage medium. According to the method, the fault prediction learning model is established by analyzing the state quantity of the network equipment, and the switch fault is predicted by utilizing the fault prediction learning model obtained by training, so that the network reliability of the datacenter is improved, the system operation stability is guaranteed, and the service downtime risk and the operation complexity are reduced.

The invention belongs to the technical field of personalized learning, and discloses a knowledge cognitive structure analysis method and system, computer equipment, a medium and a terminal, and the method comprises the steps: obtaining a joint prior feature based on a learning interaction sequence of a learner; designing a hierarchical convolutional neural network to perform spatial analysis on the learning state of the learner, and extracting spatial features including the personalized learning ability of the learner; outputting the response condition of the learner to the practice under the given heterogeneous features, and constructing the learner time-space fusion features which influence the knowledge cognitive structure and expression of the learner in the learning process; and introducing a bidirectional gate circulation unit, constructing a knowledge cognitive structure analysis model based on long-time dependence and fused spatial-temporal characteristics to dynamically diagnose the knowledge cognitive structure of the learner, and predicting the learning performance of the learner. The method is beneficial for improving the prediction precision of the knowledge cognitive structure analysis model in predicting the learning performance of the learner under the specific resources, and has certain reference significance for the development of personalized teaching.

The invention relates to a knowledge tracking method integrating a learning process and difficulty features of question knowledge points, and belongs to the technical field of education information mining. The method comprises the following steps: firstly, collecting question answering data of a learner, retaining information such as question answering starting time, question answering time, learner identification, knowledge points related to the questions, question identification and correct and wrong labels of each question in the data, preprocessing the data, and then calculating relative difficulty of each question according to the corresponding knowledge points; and then generating and storing an answer condition data sequence, and finally inputting the stored data sequence into the deep neural network model, and predicting the subsequent answer condition of the learner, namely predicting the mastering condition of the learner on the knowledge points related to the question.

The invention discloses a chip performance degradation trend prediction method based on a multi-step robust prediction learning machine, which combines an extreme learning machine and a recurrent neural network, has extremely high information fusion capability and rapid information processing capability, constructs similarity based on correlation entropy by establishing an error code book, establishes real-time prediction model updating according to chip degradation diversity and dynamics, and overcomes the influence of interference on a prediction result. Therefore, compared with the existing method, the method has higher online prediction precision, and the multi-step prediction result is more accurate compared with the existing method.

The invention provides a loan overdue prediction and learning method and device, and can be applied to the technical field of finance. An alliance learning circle is formed by combining a block chain, and a global model is formed according to each local model by utilizing the characteristics of the block chain, so that the purpose of learning updating is achieved. On one aspect, after the model training, the overdue probability of a creditor can be accurately predicted without the need for other banks to share data (the conventional loan overdue prediction must share data among banks under the condition that credit investigation is not carried out, and the loan condition of a customer to be loan in other banks is known). The overdue probability of the creditor can be accurately predicted without the need for checking the credit investigation and the like.

The invention discloses a pre-training method and device based on a Transform structure. The method comprises the following steps: firstly, acquiring an image and a video of a target object; for any segmented region in the image and the video, taking the feature symbol of the segmented region as the label of the segmented region; mask processing is carried out on the partial segmentation area of the image and the partial segmentation area of the video to obtain a first training sample and a second training sample; performing supervised prediction learning on the feature symbols of the mask region in the first training sample based on a Transform structure to obtain an initial model; initializing the pre-training model based on the initial model to obtain an initial pre-training model; and finally, performing supervised joint training on the first training sample and the second training sample by using the initial pre-training model to obtain a final pre-training model. Therefore, the model learns the spatial features and the time features of the video data at the same time, and a good pre-training model is provided for downstream tasks.

The invention discloses a method for predicting learning achievements by correcting scores of simple answers. According to the method, an answering set is determined according to a score list, a firstprofile in a first question bank is called, a second profile in a second question bank is called, and a test question adjustment coefficient of each test question and a correction score of each question and answer identifier are determined; and finally, a learning achievement of each knowledge point code is determined according to the question and answer score and the correction score. Accordingto the invention, the mastery degree of the answerer for each question and answer knowledge point is determined through the corrected test paper score, so that the learning result can be predicted more accurately. Learning achievements of a unified class or a plurality of students in the same major are summarized, and teaching quality can be evaluated.

The invention discloses a method and device for monitoring network security, an air conditioner and a household appliance. The method comprises the following steps: acquiring a security log of networking, wherein the security log records attack information that an external network attacks the networking; and processing the data in the security log based on the prediction learning model to obtain aprediction result, werein the prediction result is used for predicting to obtain abnormal information. According to the invention, the technical problem of poor network security caused by malicious attacks from an external network in networking of the air conditioning unit is solved.

The invention discloses an anticipated learning method and system for short-term time series prediction, solves the prediction problem of a short-term high-dimensional time series, and realizes accurate multi-step prediction of short-term high-dimensional data. According to the technical scheme, variables used for prediction are selected from time series data, predictive learning oriented to short-term time series prediction is conducted based on two trained neural network models, and finally the part, needing to be predicted, of the selected prediction variables is output.

The invention provides a multi-source information acquisition system for online learning student behavior analysis. The multi-source information acquisition system comprises an eye movement information acquisition system, a computer operation information acquisition system and a student face image information acquisition system; the eye movement information collection system collects eye movement information of students in the learning process; the computer operation information acquisition system comprises a learning process screen recording system, a mouse event recording system, a keyboard event recording system and a group discussion and message recording system which are respectively used for acquiring computer screen recording data, mouse event information, keyboard event information and group discussion and message information; the student face image information acquisition system acquires face image information; and the three systems perform synchronous acquisition of multi-source data by adopting a multi-thread method. According to the invention, synchronous acquisition, recording and playback of various signal data can be realized, learning behaviors of students in online courses can be analyzed, learning content difficulties of the students can be effectively mined, learning states of the students can be evaluated, and learning performance can be predicted.

The invention provides a system for predicting a learning effect based on a user online learning behavior mode. The system comprises a learning behavior information acquisition module, a learning efficiency calculation module, a learning behavior mode calculation module, a learning ability-motivation calculation module and a learning effect prediction module. The learning behavior information acquisition module correspondingly acquires user learning behavior information and user basic information. The learning efficiency calculation module generates a learning efficiency matrix of the user according to the learning behavior information of the user in different types of online courses. The learning behavior pattern calculation module generates user classification information and learning behavior pattern information of a user. The learning ability-motivation calculation module generates learning ability information and learning motivation information of the user in online course learning. And the learning effect prediction module predicts the learning effect of the user in online course learning. The invention further provides a prediction method adopting the system for predicting the learning effect of the user in different types of online courses.

The invention belongs to the field of education data mining, and provides a knowledge and skill dynamic diagnosis method oriented to space-time evolution. The method comprises the following steps of: firstly, constructing a knowledge heterogeneous graph according to resource characteristics, and then dynamically updating the knowledge and skill state of a learner in time and space dimensions, therefore, the future performance of the learner is predicted, and the knowledge mastering condition of the learner is diagnosed. According to the method, a big data technology, deep learning and a natural language processing technology are comprehensively utilized, knowledge points of a learner are modeled from time and space, the knowledge state of the learner is influenced by introducing learning features and forgetting features, and a knowledge structure of the learner is updated by providing space-time cascade operation, the knowledge and skills of the learner can be diagnosed scientifically and comprehensively, future performance of the learner can be predicted, personalized recommendation practice can be performed on knowledge points with low skill mastery, and personalized teaching can be performed on knowledge points with poor performance in the future.

The invention discloses a human-vehicle interaction habit learning method and system based on a breadth learning algorithm. Through the method provided by the invention, the learning is performed through an increment learning model according to vehicle using habit data of a user and real-time data of the user, and prediction learning content corresponding to the vehicle using habits of the user can be predicted, so that the user is unnecessary to frequently control the vehicle, but the vehicle directly executes the control according to the learning content, the interaction frequency between the vehicle and the user is reduced, and the user experience is improved.