In recent years, unmanned aerial vehicle (UAV) assisted mobile edge computing (MEC) has emerged as a novel paradigm for providing computing services to devices in the industrial Internet of Things (IIoT) domain. However, due to the constraints of a single UAV’s battery and processing capabilities, it is unable to fulfill the computational and communication requirements of IIoT devices. We presents a multi-UAV-assisted IIoT system wherein computational services for IIoT devices are collaboratively provided by a terrestrial base station (BS) and UAVs. Considering the impact of energy consumption on the environment, we formulate a long-term weighted system energy consumption minimization problem by jointly optimizing UAVs trajectories, offloading and resource allocation decisions while considering the constraints of task deadlines. Due to the network dynamics and complexity of the optimization problem, we reformulate it as a Markov decision process (MDP) and apply a multi-agent twin delayed deep deterministic policy gradient (MATD3) approach to solve the problem. The simulation results obtained show that our proposed MATD3-based method achieves 6% higher rewards and 30% faster convergence over previous algorithms such as multi-agent deep deterministic policy gradient and the proximal policy optimization.

Abstract Knowledge graph embedding aims to map entities and relations into a low‐dimensional vector space for easy manipulation. However, frequent entities are updated more often than infrequent ones during training, leading to inadequate representation of the latter's embeddings, which, in turn, affects the model's overall performance in downstream tasks. To address this issue, we propose a semantic information guide and enhance (SGE) method. The SGE tackles the heterogeneity in the frequency of entities through semantic reconstruction and a guidance network. The semantic reconstruction strengthens the semantic relevance among all entities and connects entities with different frequencies in semantic space. The guidance network extends these connections to knowledge space, enhancing the expression abilities of infrequent entities’ embeddings without compromising the embeddings of frequent entities. Experiments with four commonly used benchmark datasets show that the SGE method improves the performance of baseline models in most cases and that the method is model‐independent.

Abstract In knowledge graph embedding, multidimensional representations of entities and relations are learned in vector space. Although distance-based graph embedding methods have shown promise in link prediction, they neglect context information among the triplet components, i.e., the head_entity, relation, and tail_entity, limiting their ability to describe multivariate relation patterns and mapping properties. Such context information denotes the entity structural association inside the same triplet and implies the correlation between entities that are not directly connected. In this work, we propose a novel knowledge graph embedding model that explicitly considers context information in graph embedding via triplet component interactions (TCIE). To build connections between components and incorporate contextual information, entities and relations are represented as vectors comprised of two specialized parts, enabling comprehensive interaction. By simultaneously interacting with one-hop related head and tail entities, TCIE strengthens the connections between distant entities and enables contextual information to be transmitted across the knowledge graph. Mathematical proofs and experiments are performed to analyse the modelling ability of TCIE in knowledge graph embedding. TCIE shows a strong capacity for modelling four relation patterns (i.e., symmetry, antisymmetry, inverse, and composition) and four mapping properties (i.e., one-to-one, one-to-many, many-to-one, and many-to-many). The experimental evaluation of ogbl-wikikg2, ogbl-biokg, FB15k, and FB15k-237 shows that TCIE achieves state-of-the-art results in link prediction.

The residual network is now one of the most effective structures in deep learning, which utilizes the skip connections to “guarantee" the performance will not get worse. However, the non-convexity of the neural network makes it unclear whether the skip connections do provably improve the learning ability since the nonlinearity may create many local minima. In some previous works [Freeman and Bruna, 2016], it is shown that despite the non-convexity, the loss landscape of the two-layer ReLU network has good properties when the number m of hidden nodes is very large. In this paper, we follow this line to study the topology (sub-level sets) of the loss landscape of deep ReLU neural networks with a skip connection and theoretically prove that the skip connection network inherits the good properties of the two-layer network and skip connections can help to control the connectedness of the sub-level sets, such that any local minima worse than the global minima of some two-layer ReLU network will be very “shallow". The “depth" of these local minima are at most O(m^(η-1)/n), where n is the input dimension, η<1. This provides a theoretical explanation for the effectiveness of the skip connection in deep learning.