International Business Machines Corporation

Yubo Li

Qing Fa Wang

Chao Xue

Yonghua Lin

Guang Cheng Li

Chao Zhu

Wei Dong Wang

Junsong Wang

Ke Jin

Yanyan Hu

Jie Zhang

Shi Lei Zhang

Mingming Cai

Qi Ming Teng

Chang Xu

Rong Yan

Cheng Fang Wang

Shiwan Zhao

Hang Liu

Li Cao

Yan Gong

Disclosed are a computer-implemented method, a system, and a computer program product for system-level tunable parameter identification. Performance characteristic data for an application to be tuned can be obtained by one or more processing units. At least one system-level tunable parameter for the application to be tuned can be identified by one or more processing units based on the obtained performance characteristic data for the application to be tuned and a pattern between training performance characteristic data and a set of training system-level parameter-related correlation coefficients. The set of training system-level parameter-related correlation coefficients can be respective correlation coefficients of system-level tunable parameters with respect to at least one performance metric.

System-level tunable parameter identification

A computer-implemented method, system, and computer program product are provided for optimization with domain knowledge requirements. The method includes receiving, by a processor device, domain knowledge requirements for a target system. The method also includes defining, by a domain knowledge manager, a status of the domain knowledge employing a factor responsive to the domain knowledge. The method additionally includes computing, by an effective metric generator, effective performance metrics responsive to the status of the domain knowledge requirements and real performance metrics. The method further includes generating, by an optimization engine, a target system configuration responsive to the effective performance metrics and the real performance metrics. The method also includes improving the target system by changing a state of a function in the target system responsive to the target system configuration.

Target system optimization with domain knowledge

In some examples, a method includes receiving an artificial intelligence (AI) system scenario definition file from a user, parsing the definition file and building an application workflow graph for the AI system, and mapping the application workflow graph to an execution pipeline. In some examples, the method further includes automatically generating, from the workflow graph, application executable binary code implementing the AI system, and outputting the application executable binary code to the user. In some examples, the execution pipeline includes one or more building blocks, and the method then further includes collecting running performance of each of the building blocks of the execution pipeline in a runtime environment.

Artificial intelligence workflow builder

A method, system, and computer program product for detecting multi-scale objects. The method may include receiving a sample dataset including multi-scale objects associated with a specific environment, where the sample dataset has an existing resolution. The method may also include inputting the sample dataset into a trained neural network, where the trained neural network has a plurality of scale regions. The method may also include processing the sample dataset for detecting the multi-scale objects by means of the trained neural network. The method may also include calculating a distribution of contribution degree in a course of processing the sample dataset, where the contribution degree is associated with each of the plurality of scale regions. The method may also include generating a set of configuration parameters associated with the specific environment for the trained neural network based at least in part on the distribution of contribution degree.

Multi-scale object detection with a trained neural network

In some examples, a method includes receiving a test image from a user or client, the test image depicting an anatomical region or structure of a human body, obtaining a reference image corresponding to the anatomical region or structure depicted in the test image, and analyzing the test image and the reference image to obtain a set of differences between the two images. In some examples, the method further includes, based at least in part on the set of differences, detecting a possible abnormality in the test image and outputting a result to the user or client. In some examples, a fast R-CNN is used to detecting the possible abnormality in the test image.

Reference image guided object detection in medical image processing

Technology for performing object tracking of an object in a video image using both object detection algorithm and an object tracking algorithm. Because object detection algorithms are generally slower, but accurate, than object tracking algorithms, only certain sampled video frames are subject to the object detection algorithm. The non-sampled frames locate the object using the object tracking algorithm, using the most recent previous object detected frame as a starting frame. In this way, the object detection algorithm is used to periodically, effectively calibrate the object tracking algorithm by using a lagging detection result to hide latency introduced by the object detection algorithm.

Stream object tracking with delayed object detection

A method, computer system, and computer program product for model selection for training a new dataset is provided. The present invention may include choosing a model from a set of models to be evaluated for training the new dataset, selecting a sample input from a subset of the new dataset, calculating a model activation score for each of the sample inputs in the chosen model, calculating an accumulated model activation score for the chosen model, depending on the model activation score of each of the sample inputs in the chosen model, calculating an accumulated model activation score for each model from the set of models to be evaluated for training the new dataset, and selecting the model for training the new dataset with the highest accumulated model activation score.

Model matching and learning rate selection for fine tuning

A method comprises receiving a first received signal strength indicator (RSSI) of a first beacon in an array of beacons and receiving a second RSSI of a second beacon in an array of beacons, calculating a RSSI of the array (r) as a function of the first RSSI and the second RSSI, retrieving a calibrated RSSI value of the array (r') from a memory, determining whether r&#x3e;r&#x2032;, and outputting a signal to a user device responsive to determining that r&#x3e;r&#x2032;.

Beacon array

Examples of techniques for running kernel dependent applications in containers are disclosed. In one example implementation, a method includes initiating, by a processing system, a user mode Linux (UML) kernel in a container of a host. The method further includes initiating, by the processing system, an agent inside the UML, the agent configured to handle system call redirection from a host kernel of the host. The method further includes enabling, by the processing system, a system call hook kernel module on the host to hook a system call received from the container. The method further includes determining, by the processing system, whether the UML kernel is running inside the container. The method further includes, based at least in part on determining that the UML is running inside the container, passing, by the processing system, the system call to the UML kernel inside the container.

Running a kernel-dependent application in a container

A method, computer program product, and computer system are provided for running an application over a host kernel. The method intercepts a system call initiated by a container instance running on a host kernel, then determines whether the initiating container instance is tagged with a mapping to a virtual machine (VM) also running on the host kernel. The method further redirects the system call to the mapped VM responsive to the initiating container instance being tagged with a mapping to a VM, wherein the system call is to be processed by the VM.

Running kernel dependent applications inside a container

A method includes obtaining a first set of parameters corresponding to a first video, such parameters including a first skip number and a first segment length. The method includes obtaining a first set of image feature values corresponding to the first video and storing the first set of parameters and the first set of image feature values as reference data. The method includes calculating a second set of image feature values corresponding to a second video and comparing the second set of image feature values to the reference data, and determining, based on the comparing, that the second set of image feature values exceeds a threshold. The method includes calculating, in response to the determining, a variance. The method includes generating, based on the variance, a second set of parameters corresponding to the second video, the second set of parameters including a second skip number and a second segment length.

Dynamic sequential image processing

A resource management method, system, and computer program product for creating a dummy virtual machine (VM) in a Virtual Machine (VM) hypervisor and a dummy container in a container engine for a resource management purpose, adding hooks for each of the VM and the container, and calling the container engine to update a resource allocation in the dummy container when the hook of the VM is started and the VM hypervisor to update a resource allocation in the dummy VM when the hook of the container is started.

Growing community of inventors

Beijing, China

Yubo Li