271results about "Genetic models" patented technology

The present disclosure is related to framework for automatically and efficiently finding machine learning (ML) architectures that are optimized to one or more specified performance metrics and / or hardware platforms. This framework provides ML architectures that are applicable to specified ML domains and are optimized for specified hardware platforms in significantly less time than could be done manually and in less time than existing ML model searching techniques. Furthermore, a user interface is provided that allows a user to search for different ML architectures based on modified search parameters, such as different hardware platform aspects and / or performance metrics. Other embodiments may be described and / or claimed.

A technique related to codon optimization is provided. In a method for optimizing nucleic acid sequences expressing proteins in a host, protein subsequence-nucleic acid subsequence pairs are obtained from collected highly expressed protein sequences and their encoding nucleic acid sequences to form a training set. And training a neural machine translation model by using the training set. The neural machine translation model realizes translation from an amino acid sequence to a codon sequence. A protein sequence needing codon optimization is cut into protein subsequences. And translating a protein subsequence from an amino acid sequence to a codon sequence by using the trained neural machine translation model. The translated subsequences are superimposed to merge into a full-length codon sequence. In the process of superimposing to synthesize the codon sequence, according to the frequency or number of synonymous codons corresponding to each amino acid position of the protein sequence, the synonymous codon with the highest occurrence frequency or number is selected as the optimal codon at the position.

The present disclosure is related to framework for automatically and efficiently finding machine learning (ML) architectures that are optimized to one or more specified performance metrics and / or hardware platforms. This framework provides ML architectures that are applicable to specified ML domains and are optimized for specified hardware platforms in significantly less time than could be done manually and in less time than existing ML model searching techniques. Furthermore, a user interface is provided that allows a user to search for different ML architectures based on modified search parameters, such as different hardware platform aspects and / or performance metrics. Other embodiments may be described and / or claimed.

The present disclosure provides various devices, systems, and methods for iterative shimming in magnetic resonance imaging (MRI). In one aspect, a method of iterative shimming can include generating a field map of a magnetic field B0 from an MRI system. The MRI system can include a shim tray and an array of permanent magnets to generate the magnetic field B0. The shim tray can include slots to receive shim magnets. The method can further include shimming the MRI system by iteratively: applying the field map to a genetic algorithm to determine a set of slots to receive a set of shim magnets, installing the set of shim magnets in the set of slots, generating a next field map of the magnetic field B0, and either performing a next iteration based on the next field map or determining to not perform the next iteration based on the next field map.

The present invention provides a data center intelligent liquid cooling control method and system, which belongs to the field of liquid cooling technology, and comprises the following steps: S101: obtaining optimization constraints, and determining a flow parameter interval, a rotation speed parameter interval and a cooling parameter interval based on the optimization constraints; S102: obtaining an initial antibody population based on an artificial immune algorithm; S103: obtaining an affinity function and affinity, screening multiple random antibody feature vectors based on the affinity, and retaining multiple preferred antibody feature vectors; S104: cloning and mutating multiple preferred antibody feature vectors, obtaining multiple copies of the preferred antibody feature vectors and multiple mutated antibody feature vectors; S105: establishing a new antibody population, performing iterative optimization and adopting a mutation countermeasure strategy; S106: when the maximum number of iterative optimization times is reached, exporting the optimized feature vector with the highest affinity, and importing it into a liquid cooling control module, and the liquid cooling control module controls the cooling unit to adjust to the corresponding parameters to save energy consumption.

The invention belongs to the technical field of aero-engine performance testing, particularly relates to a physical constraint-based aero-engine state prediction model construction method and system, and aims to solve the problems of high calculation complexity and poor data quality of an existing physical information model. The method comprises the following steps: acquiring historical operation data of the aero-engine and a physical constraint rule set of engineering simplification; predicting performance parameters by adopting a deep learning model; constructing a total loss function formed by weighting a data loss item and a physical loss item to train the model; wherein the physical loss item is generated based on the deviation degree of the predicted performance parameter and the engineering simplified physical constraint rule set, and is used for replacing the complex partial differential equation constraint. According to the method, the engineering simplified physical rule is introduced, so that the calculation overhead of model training is remarkably reduced, the model is effectively guided to learn the characteristics conforming to the physical rule, and the accuracy and generalization ability of the prediction model are remarkably improved under the condition of limited data.

The invention discloses a road detection system and method based on an expressway intelligent gateway system, and relates to the technical field of intelligent traffic control, and the system comprises a road detection module and the expressway intelligent gateway system. Wherein the road detection module is used for executing a road information detection task of an expressway; the expressway intelligent gateway system is used for carrying out dynamic resource configuration on a road information detection task; the resource allocation unit is used for allocating first edge end resources of a plurality of network edge nodes to the simple charging task in real time; determining a first complex charging task from the complex charging tasks according to the allocation condition of the first side end resources, and allocating second side end resources of a plurality of network edge nodes to the first complex charging task in real time; a second complex charging task except the first complex charging task in the complex charging tasks is sent to the cloud module; and the cloud module is used for allocating cloud resources to the second complex charging task in real time to obtain a third processing result.

The invention discloses an integrated circuit fault supervision method and system for dynamic time slot allocation, and the method comprises the steps: collecting and preprocessing multi-source time slot data, and generating a standardized data set; based on the standardized data set, genetic programming is used for generating a mathematical formula for describing the relation between time slot allocation and fault detection; clustering the standardized data set by using fuzzy C-means clustering, and identifying a data state category; dynamically adjusting a time slot allocation strategy in combination with a mathematical formula and a clustering result, and generating a repair result; if abnormity is detected, an adjustment formula is quickly generated based on a mathematical formula and a clustering category, and repair is executed; and feeding back the distribution and repair result, and adaptively updating the mathematical formula and the clustering result. According to the invention, efficient adaptive optimization of dynamic time slot allocation and intelligent fault supervision in an integrated circuit system is realized.

One embodiment of the invention discloses a method and device for optimizing network anti-virus indexes. The method comprises following steps: obtaining anti-virus indexes of terminals in a network, wherein the anti-virus index of each terminal in the network comprises at least one kind of information as follows: virus infection and anti-virus capability information; optimizing the anti-virus indexes of terminals in a network by means of the genetic algorithm to obtain the optimized anti-virus indexes of terminals in the network; obtaining the optimized network anti-virus indexes based on the optimized anti-virus indexes of terminals in the network. One embodiment of the invention also discloses a device for optimizing network anti-virus indexes.

The present disclosure relates generally to a gaming system, device, and method supportive of a self-evolving, AI-based playstyle models. A gaming system, device, and method are provided that identify data associated with a gameplay session at a gaming device, provide the data to a machine learning network; receive an output from the machine learning network in response to the machine learning network processing the data using a playstyle model, the output including an indication associated with modifying the playstyle model; and modify the playstyle model based on the indication. The data includes at least one of: a gameplay decision received during the gameplay session; and an emotional state of a user in association with the gameplay decision.

The invention provides an intelligent liquid cooling control method and system for a data center, and belongs to the technical field of liquid cooling, and the method comprises the following steps: S101, obtaining an optimization constraint condition, and determining a flow parameter interval, a rotating speed parameter interval and a cooling parameter interval based on the optimization constraint condition; s102, acquiring an initial antibody population based on an artificial immune algorithm; s103, acquiring an affinity function and affinity, screening the plurality of random antibody feature vectors based on the affinity, and retaining a plurality of preferred antibody feature vectors; s104, performing cloning and mutation operation on the plurality of preferable antibody feature vectors to obtain a plurality of preferable antibody feature vector copies and a plurality of variant antibody feature vectors; s105, establishing a new antibody population, carrying out iterative optimization and adopting a mutation confrontation strategy; and S106, when the maximum number of iteration optimization times is reached, an optimization feature vector with the highest affinity is exported and imported into a liquid cooling control module, and the liquid cooling control module controls a cooling unit to be adjusted to corresponding parameters so as to save energy consumption.

The invention relates to a driving risk identification and early warning method, system and device and a medium. The method comprises the following steps: collecting a driving environment and driving data of a vehicle; fitting real-time parameters of a high-risk driving model through a genetic algorithm based on a vehicle driving environment and driving data to obtain model parameters representing driver operation characteristics and vehicle motion characteristics; based on the model parameters obtained through fitting, high-risk driving behaviors are classified and recognized through an SVM classification algorithm, and early warning is conducted on driving risks. According to the method, traffic safety is used as guidance, driving risk comprehensive evaluation and early warning under coupling influence factors are carried out by using driving data, namely, the coupling influence relationship of a driver-vehicle system, a road environment, individual characteristics of a driver and a driving state is comprehensively considered, and the driving risk is identified and early warned. The method can be widely applied to the field of driving safety.

The invention relates to the technical field of education science and technology, and discloses a self-adaptive early education training system based on cloud computing, which comprises a data acquisition module, a cloud platform data processing module, a self-adaptive analysis module, a training execution module, a feedback evaluation module and a cross-platform data synchronization interface. The multi-dimensional behavior data and the physiological indexes are fused through a transfer learning algorithm, a dynamically updated individual ability portrait is constructed, the limitation of traditional age grouping is broken through, the matching precision of a teaching scheme and the actual development stage of children is improved, and through a curriculum matching engine based on a reinforcement learning framework, the learning efficiency is improved. Teaching strategies can be dynamically adjusted according to real-time feedback, accurate adaptation of teaching content and learning progress is achieved, and the occurrence probability of advanced learning and lagged learning is reduced.

A method for performing reinforcement learning-based cargo transportation routing through a platform executed by a computing device, of embodiments, may comprise the steps of: constructing a dataset on the basis of vehicle owner information, shipper information and environment information; constructing a reinforcement learning-based route setting model from the constructed dataset; acquiring actual generated cargo information, actual environment information and actual vehicle owner information so as to provide same as an input of the reinforcement learning-based route setting model; deriving an initial cargo transportation route for each vehicle owner through the route setting model on the basis of the input; removing at least one piece of cargo information in the derived initial cargo transportation route for each vehicle owner and acquiring a reward and a penalty; and repeating the step of acquiring the reward and the penalty so as to update the route setting model, and deriving an optimal cargo transportation route for each vehicle owner.

A supportive software-based “toolbox” for optimizing finite state machine (FSM)-modeled systems. The optimization may include defining a plurality of alternative models for the FSM-modeled system and iterating one or more of the domain alternative into iterated alternatives. The optimization may include generating a plurality of stateflow representations for the FSM-modeled system according to the iterated alternatives, and based thereon, generating a model score for each of the stateflow representations predictively ranking the stateflow representations based at least in part on one or more of user experience, usability, correctness, transformability, effectiveness, and readiness.

The invention discloses a post-disaster multi-center material distribution method and system based on a double repair mechanism, and the method comprises the steps: inputting road network basic data, vehicle data, distribution point data and disaster-affected point data into a preset material distribution path planning model, and obtaining a material distribution strategy generated by the material distribution path planning model, wherein the material distribution path planning model takes the minimization of the total time of a vehicle from each distribution point to all disaster-affected points as an optimization problem, the optimization problem is solved through a memetic algorithm to obtain an optimal solution of path planning, and the memetic algorithm is used for repairing individuals with vehicle constraint violation in the initial population through a double repair mechanism. According to the post-disaster multi-center material distribution method and system based on the double-repair mechanism, the double-repair mechanism is provided, infeasible individuals in a population are repaired as much as possible, search of an algorithm can be carried out towards a feasible area, and an optimal material distribution strategy is obtained.

The embodiment of the invention relates to a processing method and device based on an ontology model, equipment, a medium and a product. The method includes creating an associated plurality of production domain instance objects related to production scheduling based on an ontology model, the plurality of production domain instance objects each having an attribute. The method further includes obtaining a first class of constraints and a second class of constraints based on attributes of the plurality of production field instance objects. Wherein the first class of constraints comprise preparation rules for production scheduling, and the second class of constraints comprise restriction rules for production scheduling. The method further includes updating the plurality of production domain instance objects by applying the first class of constraints. The method further comprises the step of executing production scheduling on the updated production field instance object by applying a second type of constraints to obtain a scheduling result. According to the embodiment of the invention, modeling and constraint of production scheduling can be easily realized, the development difficulty of a production scheduling tool is reduced, and the method has a wide application range.

We provide the system. [Solution] A means for storing and performing initial processing on communication information obtained from a data acquisition device, A detection means for analyzing the communication information and detecting anomalies, A means for generating and presenting a warning based on the detected anomaly, A self-learning method that improves the accuracy of analysis and detection using machine learning models, A means for visualizing the aforementioned communication information and warnings and for creating a report, A means of notifying mobile terminals when an anomaly is detected in the communication network, A system that includes this.

The system according to this embodiment aims to force users to perform muscle training by controlling the operation of home appliances. [Solution] The system according to the embodiment comprises a generating AI agent, a sensor, and a control unit. The generating AI agent is incorporated into home appliances and electrical products. The sensor counts the number of muscle training repetitions. The control unit controls the operation of the home appliance based on the number of muscle training repetitions counted by the sensor.

A timetable creation apparatus 100 for correcting a target timetable being a train timetable to be used to control a group of trains by using a predicted passenger demand to thereby create a new target timetable, includes: an objective function generation unit 115 that generates an objective function for an operation headway between trains included in the group of trains by using the predicted passenger demand; a constraint condition generation unit 117 that derives constraint conditions which an arrival time and a departure time of each of the trains at each of stations should satisfy for operation of the group of trains; and a candidate timetable creation unit 119 that creates a candidate timetable as an update candidate for a target timetable by using an arrival time and a departure time of each of the trains at each of the stations derived by optimizing the objective function under the constraint conditions.

We provide the system. [Solution] Means for collecting and analyzing multilingual information, A means of designing a new standard representation using a generative algorithm, means for converting the multilingual information into the standardized expression and compressing it, A means for optimizing the predictive model using the transformed information, A means of performing bidirectional communication with the user using a predicted model, A means of presenting information translated in real time, A system that includes this.

A method for identifying key targets or biomarkers of a disease at a cell specific level according to a two-step process that comprises: (1) identifying cell type-specific candidate targets / biomarkers with genetically defined causal relationships to one or more disease features by performing causal analysis (e.g., 2 sample mendelian randomization); and (2) ranking candidate targets / biomarkers by analyzing the interconnectivity between the identified candidate targets / biomarkers and common interactors / regulators in a customized cell type-specific knowledge graph.

The system according to this embodiment aims to automatically generate and provide optimal training programs tailored to the skill levels and goals of employees. [Solution] The system according to the embodiment comprises an analysis unit, a generation unit, and a monitoring unit. The analysis unit analyzes employees' historical data, current skill levels, and individual goals. The generation unit automatically generates and provides an optimal training program based on the analysis results obtained by the analysis unit. The monitoring unit monitors training progress in real time based on the training program generated by the generation unit and provides feedback and support.

The invention relates to the technical field of electric power communication, in particular to an electric power communication system circuit route planning method and system and computer equipment. The method comprises the following steps: firstly, collecting node and link data under a Network X framework to construct an initial topology, and screening out an effective network topological graph through a connected component algorithm; then, an optimal routing multi-objective optimization function of the total edge number, the total distance and the total resource utilization rate is defined based on the optimal routing multi-objective optimization function; and then, an ant colony algorithm is combined with a pruning strategy to obtain an effective route set, and a high-separation-degree route group is optimized and divided through a genetic algorithm. According to the method, multiple optimization targets can be comprehensively considered, the reasonability and the efficiency of routing planning are improved, and the reliability and the stability of an electric power communication system are enhanced.

Provided is a technique relating to codon optimization. A method for optimizing a nucleic acid sequence for expression of a protein in a host comprises: obtaining a protein subsequence-nucleic acid subsequence pair according to collected highly expressed protein sequences and encoding nucleic acid sequences thereof, so as to form a training set; using the training set to train a neural machine translation model, wherein the neural machine translation model is used to realize translation from an amino acid sequence to a codon sequence; cleaving the protein sequence requiring codon optimization into protein subsequences; using the trained neural machine translation model to translate the protein subsequences from amino acid sequences to codon sequences; and overlapping the translated subsequences, so as to combine same into a full-length codon sequence, wherein during overlapping to synthesize the codon sequence, the synonymous codon with the highest occurrence frequency or number is selected as the optimal codon for the position according to the frequency or the number of the synonymous codon corresponding to each amino acid position of the protein sequence.

The invention relates to a multi-unmanned aerial vehicle cooperation edge calculation method based on reinforcement learning. Wherein the multiple unmanned aerial vehicles serve as mobile base stations and provide calculation services for ground users; presetting the period frequency and the storage size of a central processing unit for providing calculation service on each unmanned aerial vehicle; taking task calculation service delay, energy consumption and unmanned aerial vehicle track weighted sum minimization as optimization targets; carrying out pre-training by adopting a genetic algorithm, and taking a fitness function of the genetic algorithm as a reward function of a multi-agent depth deterministic strategy gradient algorithm; determining an unloading strategy and a track of each unmanned aerial vehicle by adopting the algorithm; and a multi-intelligent depth deterministic strategy gradient algorithm is adopted to determine a delivery unmanned aerial vehicle of each task calculation result, and the result is fed back to a user. According to the method, the flight path of each unmanned aerial vehicle, the user service quality and the energy efficiency of the whole system are fully considered, timely delivery of tasks is ensured, and the whole energy consumption of the system is reduced.

Digital object library management systems and methods for machine learning applications are taught herein. Such a method includes populating a digital object library with a number of machine readable digital objects, modifying the digital objects to include additional machine readable data about the digital objects or other digital objects and the relationships among existing digital objects, generating lists of objects for use in construction and verification of machine learning models used to classify unknown objects into one or more categories, building queries to generate object lists, initiating model generation, in which a machine learning model used to classify unknown objects into one or more categories is generated, initiating model evaluation, storing models, object lists, evaluation results, and associations among these objects, generating a visual display of object metadata, lists, relational information, and evaluation results and running distributable algorithms across the library of digital objects.

The application discloses a layout planning method based on reinforcement learning and a genetic algorithm, and comprises the following steps: 1, generating an original population of the genetic algorithm through perturbation B*‑tree; 2, calculating the fitness of all individuals, and selecting an individual with the highest fitness value as the optimal individual in the population; 3, according to the probability, selecting whether to use a crossover operator in an iteration, if the crossover operator is selected, selecting two parent individuals from the population, and generating offspring individuals through the crossover; 4, according to the probability, selecting whether to use a mutation operator in the iteration, if the mutation operator is selected, selecting a parent individual from the population, and generating an offspring individual through the mutation; 5, optimizing the offspring individual by using a reinforcement learning intelligent agent, and then determining whether to replace the parent individual according to the fitness of the offspring individual; and 6, the iteration is ended, and the optimal individual is obtained. The application efficiently reduces the area and line length of layout planning.

The present invention relates to a distribution network multidimensional measurement configuration method and system for improving risk perception capabilities, which includes: analyzing data requirements for the types of disasters faced by the distribution network in the region; selecting distribution network multidimensional measurement equipment based on the analysis results of the data requirements and taking into account the micro-meteorological factors in the region; considering the real-time data transmitted by the multidimensional measurement equipment configured in the distribution network, comparing the line failure probability results obtained based on the distribution network disaster mechanism model with the actual line failure probability, and adopting a distribution network risk perception accuracy improvement method; improving the distribution network risk perception capability, ensuring that the measurement equipment configuration has good economic efficiency as the goal, and taking into account the observability of important load nodes, using a genetic algorithm to solve the number and location of the multidimensional measurement configuration in the distribution network. The present invention achieves the improvement of the distribution network disaster loss identification capability in the event of failure of the electrical measurement device, and can be widely used in the field of electrical automation.