30results about How to "Fix cold start" patented technology

The invention discloses a recommendation method and system integrating user behaviors and object content. The recommendation method integrating the user behaviors and the object content comprises the steps that through correlation between behavior data of each user and content data of all the objects, a list of user interest of each user to all the objects is obtained; the similarity of the list of user interest and the content data of all the objects is calculated, the recommendation object weight of each user to the content data of each object is obtained, and the objects recommended to each user are obtained through ranking of the recommendation object weights. According to the recommendation method and system integrating the user behaviors and the object content, the main problems that how an existing recommendation conducts cold start and how the existing recommendation smoothly transmits to the normal operation state from cold start are solved, the accuracy, the coverage rate, the novelty and the like of the system are greatly improved, the accuracy and the personalization of the recommendation system can be well improved, and more users are attracted and use the recommendation system.

The invention discloses a position recommendation method which comprises the following steps: S0, acquiring all position data, user historical behavior data and a position currently browsed by a target user, and acquiring a position browsing sequence of each user according to all the user historical behavior data; S1, generating a new position sequence group by using the position browsing sequence of each user in combination with a random walk algorithm; S2, training a Word2vec model by adopting a Skip-gram framework to generate a feature vector of each position; and S3, calculating feature vector cosine similarity between the position currently browsed by the target user and all positions, and forming a position candidate set. The position candidate set is obtained as a position recommendation set by using the feature vector cosine similarity between the position currently browsed by the target user and all the positions, and the position recommendation method is high in real-time performance and accuracy, and avoids the problems of Matthew effect and cold start.

The invention provides a novel automobile thermal insulating heater, including a thermal insulating water tank, a sensor, a water pump, an electromagnetical valve, a pipeline and a control display. The working principle is as follows: when the key of the automobile is off and the engine stops, the control display controls the electromagnetical valve to open, the water pump runs, and after high-temperature cooling fluid in the engine cooling system enters the thermal insulating water tank under the action of the water pump, the electromagnetical valve and the water pump are closed and heat preservation is carried out to the cooling fluid in the thermal insulating water tank; when the key of the vehicle is on and the engine can be started, the control display controls the electromagnetical valve to open, the water pump runs counterclockwise and after the high-temperature cooling fluid in the thermal insulating water tank enters the engine cooling system for preheating, the engine under the action of the water pump, the electromagnetical valve and water pump are closed. At that time, cold starting of the engine can be avoided when the vehicle engine are restarted and additionally after the engine starts, the cooling fluid in the engine cooling system quickly reaches normal temperature for the working requirement of the engine. Compared with the prior art, the invention has the advantages of improving the cold starting performance of the engine, reducing the abrasion of the engine, saving energy and decreasing pollution and the like, therefore, the thermal insulating heater has good usage and popularization valve.

A lithium-ion battery thermal management method based on movable fins and phase-change materials, which can perform thermal management on its operation according to different operating temperatures of lithium-ion batteries: that is, when a lithium-ion battery works at a medium temperature, the battery temperature reaches the temperature of the phase-change material. Melting point, the phase change material converts the heat of the battery into its own latent heat storage through solid-liquid phase transition; when the phase change material is fully liquefied, the thermal management system drives the fins to make contact with the surface of the battery cell, conducts heat through the fins and Refrigerant convection, supplemented by liquid phase-change material to absorb heat for high-temperature heat dissipation; when the battery stops working and the battery temperature begins to drop and approaches the melting point of the phase-change material, the thermal management system retracts the fins to separate them from the surface of the battery cell , which slows down the external dissipation of the phase change material and the internal heat of the battery, and prevents the battery pack from dropping too quickly in a low temperature environment. This design can effectively ensure that the lithium-ion battery can obtain an ideal operating temperature and surface temperature difference under different operating temperatures.

The invention provides a method, device and terminal for accelerating starting of web page based on object pool. The method comprises the steps of before a web page is loaded, pre-building a web page environment object, submitting the web page environment object to a object pool for integral management to achieve the direct opening of the web page through the web page environment object based on a universal resource locator, thus during cold starting, saving the time needed to initialize the web page running environment and during hot starting ,saving time to load variable environment to accelerate the web page loading speed, and enhance user's experience. Moreover, according to the method, device and terminal for accelerating starting of web page based on object pool, the object pool manages one or more web page environment objects, each web page environment object takes memory space, the technical scheme automatically adjusts the volume of the object pool based on the content volume to avoid the system burden caused by the memory space occupation by the object pool.

The invention relates to a parking heating system for a motor vehicle driven by an electric motor and a method for implementing a parking heating function in a motor vehicle with an electric drive train.

The invention relates to a cloud game accelerated starting method and device and electronic equipment. The method comprises the following steps of: when a cloud game process corresponding to any cloud game running in a cloud host is terminated, performing process keep-alive processing on the cloud game process based on a process keep-alive mechanism on the cloud host, restarting the cloud game process on the cloud host, and switching the cloud game process to a background for running; receiving a connection request which is triggered by a game process on a user terminal and aims at the cloud game, the connection request comprising a starting parameter used for waking up the cloud game process; and in response to the connection request, transmitting the starting parameter to the cloud game process, and switching the cloud game process to a foreground for operation.

The invention discloses a content recommendation method based on a heterogeneous feature deep residual network, and the method comprises the steps: obtaining the source data of to-be-recommended content, carrying out the data conversion of the source data, and obtaining weighted hybrid embedded data; obtaining predicted score data according to the weighted hybrid embedded data and a deep residualnetwork model; and obtaining a recommendation result of the content according to the prediction score data. According to the embodiment of the invention, the source data of the domain content is processed, the processed data is input into the residual network model to obtain the predicted score data, and then the domain content is accurately recommended according to the predicted score data, so that the calculation method is high in efficiency, low in resource occupancy. And meanwhile, the diversified data can also avoid cold start when new users are recommended.

The invention discloses a rating prediction method for a joint topic model and a heterogeneous information network. The steps of the present invention are as follows: step (1) for a designated user commodity pair, use a topic model to extract comment information, thereby constructing a vector representation of the user and commodity; step (2) construct a heterogeneous information network by using commodity attribute information and user common purchase information; Step (3) Extract the final relationship representation vector of the user-product pair from the heterogeneous network; Step (4) For the user-product pair, connect the user vector, relationship representation vector, and product vector representation, and input it to AFM to achieve score prediction; step 5. According to the predicted scoring data calculated by the model and the real scoring data, calculate the RMSE value, and use this value as the evaluation index of the model effect. The invention solves the problems of cold start, poor explainability and low accuracy in score prediction.

The present invention discloses a method and device for speech recognition. The method includes: step 101, acquiring the speech recognition models of multiple users; step 102, performing statistics on the speech recognition models of the multiple users, and determining that the similarity is greater than N user sets equal to the first threshold, where N is an integer greater than or equal to; step 103, perform corpus training on the speech recognition models of the i-th user set to generate the i-th user set speech recognition model, 1≤i≤N. Through the method of the present invention, the commonality of similar users can be used to quickly accumulate user-level corpus, quickly build a user-level language model, solve the problem of poor support for user-level speech recognition, and avoid cold starts caused by too few user sentences, providing Better user-level speech recognition effect improves user experience.

The invention relates to the technical field of engine emission control, in particular to a device, method and automobile for simultaneously preheating an engine and a catalytic converter. The monitor, the circulation loop includes the engine, the catalytic converter and the heater, the engine and the catalytic converter are connected through a pipeline assembly, the heater is arranged between the engine and the catalytic converter, and the heater is used for controlling The engine and the catalytic converter are heated at the same time; the invention avoids the cold start of the engine by preheating the engine and the catalytic converter before the engine starts, making the engine easier to start, and reducing the fuel consumption and emission of the engine starting stage. It is greatly improved, and at the same time, the damage to the supercharger vane is reduced.

The invention discloses a recommendation method and system integrating user behaviors and object content. The recommendation method integrating the user behaviors and the object content comprises the steps that through correlation between behavior data of each user and content data of all the objects, a list of user interest of each user to all the objects is obtained; the similarity of the list of user interest and the content data of all the objects is calculated, the recommendation object weight of each user to the content data of each object is obtained, and the objects recommended to each user are obtained through ranking of the recommendation object weights. According to the recommendation method and system integrating the user behaviors and the object content, the main problems that how an existing recommendation conducts cold start and how the existing recommendation smoothly transmits to the normal operation state from cold start are solved, the accuracy, the coverage rate, the novelty and the like of the system are greatly improved, the accuracy and the personalization of the recommendation system can be well improved, and more users are attracted and use the recommendation system.