Abstract: A time-aware application task scheduling system for a green data center (GDC) that includes a task scheduling processor coupled to one or more queue processors and an energy collecting processor connected to one or more renewable energy sources and a grid power source. The systems is capable of determining a service rate for a plurality of servers to process a plurality of application tasks in the GDC and scheduling, via processing circuitry, one or more of the application tasks to be executed in one or more of the servers at a rate according to a difference in an accumulated arriving rate for the plurality of application tasks into the one or more queues and a removal rate for the plurality of application tasks from the one or more queues. The system is further capable of removing the one or more application tasks from their associated queues for execution in the scheduled one or more servers.

Abstract: An AUC-maximized high-accuracy classification method and system for imbalanced datasets integrates an under-sampling-and-ensemble strategy, a true-outliers-removing strategy and a fake-outliers-concealing strategy, with the hope to effectively and robustly enhance both the AUC and the accuracy metrics in imbalanced classification. Applying under-sampling to construct multiple sub-datasets and assembling classification results of multiple classifiers greatly decline the risk of misclassification and lead to highly accurate and robust results in imbalanced classification task. Moreover, this invention pays attention to detect and identify extremely hidden outliers in a sub-dataset which includes a sub-majority dataset and the entire minority dataset. In this way, more hidden outliers can be located and thus exert less influence on the decision boundary, which contributes to both high AUC and accuracy.

Abstract: A method of scheduling tasks includes receiving a plurality of application tasks into one or more queues of a green data center (GDC), and determining a service rate for a plurality of servers to process the plurality of application tasks in the GDC, via processing circuitry. The method also includes scheduling, via the processing circuitry, one or more of the application tasks to be executed in one or more of the servers at a rate according to a difference in an accumulated arriving rate for the plurality of application tasks into the one or more queues and a removal rate for the plurality of application tasks from the one or more queues; and removing the one or more application tasks from their associated queues for execution in the scheduled one or more servers.

Abstract: A time-aware method of scheduling application tasks in a green data center (GDC) using a task scheduling processor and an energy collecting processor connected to one or more renewable energy sources and a grid power source. The method includes receiving energy data regarding available energy from renewable energy resources such that the renewable energy source is used first then receiving the application tasks and determining a service rate for servers to process the application tasks by using a service rate obtained by solving a profit maximization problem for the GDC by prioritizing and immediately scheduling or removing certain application tasks. The method also includes determining an initial fitness level of each representative application task based on total revenue of the GDC, an energy cost of the GDC, and a penalty associated with violated constraints.

Abstract: In the field of computer vision, without sufficient labeled images, it is challenging to train an accurate model. But through visual adaptation from source to target domains, a relevant labeled dataset can help solve such problem. Many methods apply adversarial learning to diminish cross-domain distribution difference. They are able to greatly enhance the performance on target classification tasks. GAN (Generative Adversarial Networks) loss is widely used in adversarial adaptation learning methods to reduce a across-domain distribution difference. However, it becomes difficult to decline such distribution difference if generator or discriminator in GAN fails to work as expected and degrades its performance. To solve such cross-domain classification problems, an adaptation algorithm and system called as Generative Adversarial Distribution Matching (GADM) is implemented.

Abstract: Soft sensing of nonlinear and multimode industrial processes given a limited number of labeled data samples is disclosed. Methods include a semi-supervised probabilistic density-based regression approach, called Semi-supervised Weighted Gaussian Regression (SWGR). In SWGR, different weights are assigned to each training sample based on their similarities to a query sample. Then a local weighted Gaussian density is built for capturing the joint probability of historical samples around the query sample. The training process of parameters in SWGR incorporates both labeled and unlabeled data samples via a maximum likelihood estimation algorithm. In this way, the soft sensor model is able to approximate the nonlinear mechanics of input and output variables and remedy the insufficiency of labeled samples. At last, the output prediction as well as the uncertainty of prediction can be obtained by the conditional distribution.

Abstract: A system and method of scheduling tasks include receiving inputted data task variables for a private CDC and a plurality of public clouds; initializing parameters for a plurality of representative entities in a cluster of entities; determining a fitness level of each representative entity in the cluster of entities; updating one or more task scheduling parameters for a given number of time slots based on the parameters for the plurality of representative entities and the fitness level of each representative entity; determining a total number of data tasks to be dispatched to the private CDC and the plurality of public clouds based on an iteration result of a final time slot for the given number of time slots for a global best position; and updating the data task variables using the total number of data tasks to be dispatched.

Abstract: A method of scheduling tasks includes receiving a plurality of application tasks info one or more queues of a green data center (GDC), and determining a service rate For a plurality of servers to process the plurality of application tasks in the GDC, via processing circuitry. The method also includes scheduling, via the processing circuitry, one or more of the application tasks to be executed in one or more of the servers at a rate according to a difference in an accumulated arriving rate for the plurality of application tasks into the one or more queues and a removal rate for the plurality of application tasks from the one or more queues; and removing the one or more application tasks from their associated queues For execution in the scheduled one or more servers.

Abstract: A system and method of scheduling tasks include receiving inputted data task variables for a private CDC and a plurality of public clouds; initializing parameters for a plurality of representative entities in a cluster of entities; determining a fitness level of each representative entity in the cluster of entities; updating one or more task scheduling parameters for a given number of time slots based on the parameters for the plurality of representative entities and the fitness level of each representative entity; determining a total number of data tasks to be dispatched to the private CDC and the plurality of public clouds based on an iteration result of a final time slot for the given number of time slots for a global best position; and updating the data task variables using the total number of data tasks to be dispatched.