International Business Machines Corporation

Telefonaktiebolaget Lm Ericsson (publ)

Lin Dong

Chao Xue

HuaiShan Zhang

Tao Liu

Ke Wei Sun

Yonggang Hu

Ning Sun

Xiangfang Li

Jun Feng Liu

Shuguang Yang

A computer-implemented method for training a machine learning model includes receiving a first training job at a processing device of a computer system having a plurality of graphics processing unit (GPU) resources, the first training job being part of a set of training jobs, and determining an amount of available memory in each GPU resource of the plurality of GPU resources. The method also includes loading the training job into one or more GPU resources with at least one second training job. The loading includes determining a cost model indicating an efficiency cost of each of a plurality of packing patterns, and packing the first training job and the second training job into the one or more GPU resources according to a packing pattern associated with a lowest efficiency cost.

Graphics processing unit training job allocation

An air volume control valve capable of accurately measuring an air volume is provided. The air volume control valve includes a round-shaped valve body, and a valve blade and an orifice plate provided in the round-shaped valve body, where the orifice plate is located upstream of the valve blade; an upstream pressure tapping port and a downstream pressure tapping port are respectively provided upstream and downstream of the orifice plate; a distance from both the upstream pressure tapping port and the downstream pressure tapping port to the orifice plate is 0.5 times a valve diameter; the valve blade is located downstream of the downstream pressure tapping port; an inlet side of the round-shaped valve body is provided with a convergent duct fitting; and the convergent duct fitting is located at a distance of once the valve diameter upstream of the orifice plate and is tapered along an air flow direction.

Air volume control valve capable of accurately measuring air volume

A method, system, and computer program product for hyper-parameter determination. In a method, a network architecture of a learning model may be determined, and the learning model may be configured for performing a computing task based on machine learning. A metric value record associated with a group of hyper-parameters may be obtained during hyper-parameter determination for the learning model. An estimation of a metric value may be obtained based on the network architecture, and the metric value record and an association relationship representing an association between network architectures and metric values for the network architectures. The group of hyper-parameters may be selected in response to the estimation of the metric value meeting a predefined criterion. With these embodiments, a group of hyper-parameters may be selected, and further the learning model may be trained based on the selected group of hyper-parameters.

Hyper-parameter management

Aspects of the invention include systems and methods to obtain meta features of a dataset for training in a deep learning application. A method includes selecting an initial search space that defines a type of deep learning architecture representation that specifies hyperparameters for two or more neural network architectures. The method also includes applying a search strategy to the initial search space. One of the two or more neural network architectures are selected based on a result of an evaluation according to the search strategy. A new search space is generated with new hyperparameters using an evolutionary algorithm and a mutation type that defines one or more changes in the hyperparameters specified by the initial search space, and, based on the mutation type, the new hyperparameters are applied to the one of the two or more neural networks or the search strategy is applied to the new search space.

Search space exploration for deep learning

The present disclosure relates to VoWLAN call handover solutions. In one embodiment the method comprising: making a voice call through a Wireless Local Area Network (WLAN) under whose coverage the UE is located, with a Packet Data Network (PDN) address obtained from a PDN to which the WLAN is connected; determining that a signal quality of the WLAN is not suitable for the voice call anymore; and sending the PDN address to a base station of a mobile communication network covering the UE, for facilitating a core network node of the mobile communication network to perform voice call handover from WLAN to the mobile communication network. The present disclosure also provides a UE and a core network node for implementing the same.

VoWLAN call handover method, UE and core network node

A method and a system for implementing a MBMS counting in a wireless communication network environment, wherein a plurality of User Equipments and at least one Radio Network Controller are located in the wireless communication network. The method comprises: confirming a counting response from the UE by means of sending a first message containing a Service ID list and a counting sequence number (CSN) to the UE when the CRNC receives a connection request from the UE; storing the CSN in the UE and starting a timer Tcsn configured in the UE when the UE receives the first message; not responding to a counting when a further CSN and a further Service ID list in a further message received from the CRNC are the same as the CSN and the Service ID list stored in UE and the timer Tcsn does not expire; starting a new counting response by responding to the counting and stopping the timer Tcsn when either the further CSN or the Service ID list in the further message is distinguishing from the CSN and the Service ID list stored in UE and when the timer Tcsn does not expire. According to the method and system of the present invention, the UTRAN radio resource is saved and the UEs consume less power. Meanwhile, the interferences to other users caused by the RRC connection can be avoided.

Growing community of inventors

Beijing, China

Lin Dong