Learning how to predict future events from patterns of past events is difficult when the set of possible event types is large. Training an unrestricted neural model might overfit to spurious patterns. To exploit domain-specific knowledge of how past events might affect an event’s present probability, we propose using a temporal deductive database to track structured facts over time. Rules serve to prove facts from other facts and from past events. Each fact has a time-varying state—a vector computed by a neural net whose topology is determined by the fact’s provenance, including its experience of past events. The possible event types at any time are given by special facts, whose probabilities are neurally modeled alongside their states. In both synthetic and real-world domains, we show that neural probabilistic models derived from concise Datalog programs improve prediction by encoding appropriate domain knowledge in their architecture.

In this paper, we study a joint energy harvesting (EH) and task offloading (TO) design for an energy- constrained fog computing network, which consists of a mobile terminal and two fog nodes. The energy- constrained terminal employs time switching (TS) protocol to harvest energy from the signals sent by the circuit-powered fog node, and then exploits the harvested energy to perform local computing and offload computing. Our aim is to minimize the product of energy consumption and time delay with the constraints of TS ratio and EH requirements. To determine the optimal solution of the original non-convex problem, we decouple it into three subproblems based on the time delay assumption and then solve these subproblems through our proposed constraints activation algorithm. Furthermore, we also derive the closed form expressions for the optimal TO and TS ratios. Simulation results are presented to illustrate the effectiveness and superior performance of the proposed joint design against other conventional schemes in the literature.

Existing max-margin matrix factorization (M3F) methods either are computationally inefficient or need a model selection procedure to determine the number of latent factors. In this paper we present a probabilistic M3F model that admits a highly efficient Gibbs sampling algorithm through data augmentation. We further extend our approach to incorporate Bayesian nonparametrics and build accordingly a truncation-free nonparametric M3F model where the number of latent factors is literally unbounded and inferred from data. Empirical studies on two large real-world data sets verify the efficacy of our proposed methods.

We present a probabilistic formulation of max-margin matrix factorization and build accordingly a nonparametric Bayesian model which automatically resolves the unknown number of latent factors. Our work demonstrates a successful example that integrates Bayesian nonparametrics and max-margin learning, which are conventionally two separate paradigms and enjoy complementary advantages. We develop an efcient variational algorithm for posterior inference, and our extensive empirical studies on large-scale MovieLens and EachMovie data sets appear to justify the aforementioned dual advantages.

Electricity demand is depend closely with national economic growth since the all most economical activities have to use electricity. So, it is necessary to study economic growth for electricity demand and balancing electric power supply-demand There are many semi-structured and unstructured problems in studying electric power supply-demand, which are very difficult to be solved by the traditional methods. In this paper, a laboratory of power supply-demand based on intelligent engineering is developed to do the national policy study and simulate economic operating for analyzing electricity supply-demand. It is in the closed-up model as ¿forecast - early warning - intelligent simulation experiment - expert discussion - forecast¿. The main functions of the laboratory and their features are suitable for policymakers to see what will be happened on economic operation and electricity demand with some national policies on promoting economic growth and energy efficiency An experiment about policy simulation is given to show how to use the laboratory. It has been used in 31 area and provincial electric power grids under State Grid Co. of China. The applications have showed that it is a useful tool to do policy analysis, national economic-energy simulation, and energy efficiency for both power supply and demand.

This paper considers the flocking problem of a group of autonomous agents moving in the space with a virtual leader. We investigate the dynamic properties of the group for the case where the state of the virtual leader may be time-varying and the topology of the neighboring relations between agents is dynamic. To track such a leader, we introduce a set of switching control laws that enable the entire group to generate the desired stable flocking motion. The control law acting on each agent relies on the state information of its neighboring agents and the external reference signal (or "virtual leader"). Then we prove that, if the acceleration input of the virtual leader is known, then each agent can follow the virtual leader, and moreover, the convergence rate of the center of mass (CoM) can be estimated; if the acceleration input is unknown, then the velocities of all agents asymptotically approach the velocity of the CoM, and thus the flocking motion can be obtained, however in this case, the final velocity of the group may not be equal to the desired velocity. Numerical simulations are worked out to further illustrate our theoretical results

In this paper, we consider a group of mobile agents moving in the space with point mass dynamics. We investigate the dynamic properties of the group for the case where the topology of the neighboring relations between agents varies with time. Under the assumption that the neighboring graph is always connected, we show that stable flocking motion can be achieved by using a set of switching control laws. The control laws are a combination of attractive/repulsive and alignment forces. Using the control laws, all agent velocities become asymptotically the same, collisions can be avoided between the agents, and the final tight formation minimizes all agent potentials. Moreover, we show that the velocity of the center of mass is invariant and is equal to the final common velocity. Subsequently, we study the motion of the group when the velocity damping is taken into account. In this case, we can appropriately modify the control laws to generate the same stable flocking motion. Finally, we provide some numerical simulations to further illustrate our theoretical results.

This paper considers a group of mobile autonomous agents moving in the space with point mass dynamics and with asymmetric coupling matrix. We investigate the dynamic properties of the group for the case that the topology of the neighboring relations between agents varies with time. Under the assumption that the neighboring graphs are always connected, we show that stable flocking motion can be achieved by using a set of switching control laws. The control laws are a combination of attractive/repulsive and alignment forces. By using the control laws, all agent velocities become asymptotically the same, collisions can be avoided between the agents, and the final tight formation minimizes all agent potentials. Moreover, we show that the velocity of the center of mass is invariant and is equal to the final common velocity. Finally, we study the motion of the group when the velocity damping is taken into account. We prove that the common velocity asymptotically approaches zero. In this case, we can properly modify the control laws to generate the same stable flocking motion.