In times of the Internet of Everything (IoE), the power of the Internet is growing exponentially, followed by a surge in the number of network requests. The conflict between people’s high requirements for the Quality of Experience (QoE) and limited computing resources are becoming increasingly prominent. Therefore, an appropriate offloading method is required to better ease this conflict. In this article, a highly efficient scheduling architecture of information processing under the big data flow of the IoE is proposed to enhance the scheduling performance. First, we construct a dual-channel processing model to describe the entire data flow and node devices. Second, we carefully consider the choice of the weighting method to better find a balance between dual objectives. Third, a dual-objective deep$Q$-network (DQN)-based offloading algorithm with principal component analysis weighting method (D2OP) is proposed to collaboratively minimize task response time and machine load in a more reasonable allocation. To verify the performance of the D2OP, a series of experiments are conducted from multiple angles. The experimental results demonstrate its better performance than the three comparison algorithms in reducing response time, load balance, and increasing task success ratio.

With cloud computing facing higher levels of Big Data than ever, the processor scale is rapidly expanding. Large clusters place a heavy burden on cloud service providers and the environment. High energy consumption decreases the economic benefits of cloud service providers while enormous power demands pressure on the environment. The dynamic consolidation of virtual machines (VMs), which uses live migration technology to optimize resource usage and reduce energy consumption, is sufficient for saving energy while ensuring high performance with the desired level of quality of service (QoS) between cloud providers and users. In this article, we propose a novel machine-learning algorithm called deep-Q neural network VM consolidation (DQNVMC) that combines the Q-leaning approach with deep learning neural network to find an approximately optimal solution. Furthermore, based on the real workload trace in the cloud environment, the experiments show that DQNVMC effectively reduces energy consumption while meeting the high performance of QoS requirements.

Body orientation estimation provides crucial visual cues in many applications, including robotics and autonomous driving. It is particularly desirable when 3-D pose estimation is difficult to infer due to poor image resolution, occlusion or indistinguishable body parts. We present COCO-MEBOW (Monocular Estimation of Body Orientation in the Wild), a new large-scale dataset for orientation estimation from a single in-the-wild image. The body-orientation labels for around 130K human bodies within 55K images from the COCO dataset have been collected using an efficient and high-precision annotation pipeline. We also validated the benefits of the dataset. First, we show that our dataset can substantially improve the performance and the robustness of a human body orientation estimation model, the development of which was previously limited by the scale and diversity of the available training data. Additionally, we present a novel triple-source solution for 3-D human pose estimation, where 3-D pose labels, 2-D pose labels, and our body-orientation labels are all used in joint training. Our model significantly outperforms state-of-the-art dual-source solutions for monocular 3-D human pose estimation, where training only uses 3-D pose labels and 2-D pose labels. This substantiates an important advantage of MEBOW for 3-D human pose estimation, which is particularly appealing because the per-instance labeling cost for body orientations is far less than that for 3-D poses. The work demonstrates high potential of MEBOW in addressing real-world challenges involving understanding human behaviors. Further information of this work is available at https://chenyanwu.github.io/MEBOW/.