In the current study, a novel network model is proposed for text classification based on Graph Attention Networks (GATs) and sentence-transformer embeddings. Most existing methods with a pretraining model as an input layer still treat words as the minimum processing unit. However, word embedding is not an efficient and appropriate solution when dealing with long texts containing many professional words. This study aims to design a model capable of handling text classification tasks at multilevel semantic segmentation. The main contribution of this study is that a novel GAT variant is designed using global nodes and Squeeze-and-Excitation Networks (SENet) to capture semantic information. Moreover, a novel unidirectional attention mechanism is introduced for our model to avoid the message passing of irrelevant noisy information within global nodes. The numerical results show that according to the characteristics of datasets, specific semantic information combinations can effectively improve the accuracy and performance of text classification. Without fine-tuning of the pretrained encoder, the new state-of-the-art performance is achieved on three benchmark datasets. In addition, a comprehensive analysis of the graph attention mechanism in the model for specific cases suggests that the unidirectional attention mechanism and the use of global nodes are key contributing factors to multilevel semantic fusion.

In recent years a vast amount of visual content has been generated and shared from many fields, such as social media platforms, medical imaging, and robotics. This abundance of content creation and sharing has introduced new challenges, particularly that of searching databases for similar content - Content Based Image Retrieval (CBIR) - a long-established research area in which improved efficiency and accuracy are needed for real-time retrieval. Artificial intelligence has made progress in CBIR and has significantly facilitated the process of instance search. In this survey we review recent instance retrieval works that are developed based on deep learning algorithms and techniques, with the survey organized by deep feature extraction, feature embedding and aggregation methods, and network fine-tuning strategies. Our survey considers a wide variety of recent methods, whereby we identify milestone work, reveal connections among various methods and present the commonly used benchmarks, evaluation results, common challenges, and propose promising future directions.

Fine-grained image retrieval has been extensively explored in a zero-shot manner. A deep model is trained on the seen part and then evaluated the generalization performance on the unseen part. However, this setting is infeasible for many real-world applications since (1) the retrieval dataset can be non-fixed so that new data are added constantly, and (2) data samples of the seen categories are also common in practice and are important for evaluation. In this paper, we explore lifelong fine-grained image retrieval (LFGIR), which learns continuously on a sequence of new tasks with data from different datasets. We first use knowledge distillation to minimize catastrophic forgetting on old tasks. Training continuously on different datasets causes large domain shifts between the old and new tasks while image retrieval is sensitive to even small shifts in the features. This tends to weaken the effectiveness of knowledge distillation by the frozen teacher. To mitigate the impact of domain shifts, we use the network inversion method to generate images of the old tasks. In addition, we design an on-the-fly teacher which transfers knowledge captured on a new task to the student to improve better generalization performance, thereby achieving a better balance between old and new tasks in the end. We name the whole framework as Dual Knowledge Distillation (DKD), whose efficacy is demonstrated by extensive experimental results on sequential tasks including seven datasets.

Deep learning for fine-grained image retrieval in an incremental context is less investigated. In this paper, we explore this task to realize the model’s continuous retrieval ability. That means, the model enables to perform well on new incoming data and reduce forgetting of the knowledge learned on preceding old tasks. For this purpose, we distill semantic correlations knowledge among the representations extracted from the new data only so as to regularize the parameters updates using the teacher-student framework. In particular, for the case of learning multiple tasks sequentially, aside from the correlations distilled from the penultimate model, we estimate the representations for all prior models and further their semantic correlations by using the representations extracted from the new data. To this end, the estimated correlations are used as an additional regularization and further prevent catastrophic forgetting over all previous tasks, and it is unnecessary to save the stream of models trained on these tasks. Extensive experiments demonstrate that the proposed method performs favorably for retaining performance on the already-trained old tasks and achieving good accuracy on the current task when new data are added at once or sequentially.

The focus of this survey is on the analysis of two modalities of multimodal deep learning: image and text. Unlike classic reviews of deep learning where monomodal image classifiers such as VGG, ResNet and Inception module are central topics, this paper will examine recent multimodal deep models and structures, including auto-encoders, generative adversarial nets and their variants. These models go beyond the simple image classifiers in which they can do uni-directional (e.g. image captioning, image generation) and bi-directional (e.g. cross-modal retrieval, visual question answering) multimodal tasks. Besides, we analyze two aspects of the challenge in terms of better content understanding in deep multimodal applications. We then introduce current ideas and trends in deep multimodal feature learning, such as feature embedding approaches and objective function design, which are crucial in overcoming the aforementioned challenges. Finally, we include several promising directions for future research.

Behavioral modeling is a pivot modeling issue in combat system effectiveness simulation. There are different aspects of combat behaviors which need to be modeled with multiple modeling and simulation formalisms. In this paper, we firstly analyze the characteristics of different combat behavior aspects and the choice of formalisms to cope with a specific behavioral aspect. Then we propose a state-centered multi-formalism behavioral modeling method. We also build a metamodel and a modeling environment to provide infrastructure support for this method. Finally a case study on missile behavior modeling is given. The proposed method can enable the modeler to describe different aspects of the combat behaviors appropriately by using suitable formalisms, and thus support effectiveness simulation under the background of system-of-systems counterwork.

Military simulations, especially those for personnel training and equipment effectiveness analysis, require proper human behavior models (HBMs) to play blue or red. Traditionally, the HBMs are controlled through rule based scripts. However, the doctrine-driven behavior is rigid and predictable, and more often than not unable to adapt to new situations. In most cases, the subject matter experts (SMEs) review, re-design a large amount of HBM scripts for new scenarios or training tasks, which is challenging and time-consuming. Therefore, a study of using Grammatical Evolution (GE) to generate adaptive HBMs for air combat simulation is conducted in this work. Expert knowledge is encoded with modular behavior trees (BTs) for the compatibility with the operators in genetic algorithm (GA). GE maps HBMs represented with BTs to binary strings, and uses GA to evolve HBMs with the performance fed back from simulation. Beyond visual range air combat experiments between adaptive HBMs and none-adaptive baseline HBMs are conducted to study the evolutionary process. The experimental results show that the GE is an efficient framework to generate adaptive HBMs in BTs formalism and evolve them with GA.

Decision modeling is an essential part of the combat system effectiveness simulation (CoSES), which needs to cope with the cognitive quality, diversity, flexibility, and higher abstraction of decision making. In this paper, a multi-paradigm decision modeling framework is proposed to support decision modeling at three levels of abstraction based on domain-specific modeling (DSM). This framework designs a domain-specific modeling language (DSML) for decision modeling to raise the abstraction level of modeling, transforms the domain-specific models to formalism-based models to enable formal analysis and early verification and validation, and implements the semantics of the DSML based on a Python scripts framework which incorporates the decision model into the whole simulation system. The case study shows that the proposed approach incorporates domain expertise and facilitates domain modeler’s participation in CoSES to formulate the problem using DSML in the problem domain, and enables formal analysis and automatic implementation of the decision model in the solution domain.

The development of computer processor has stepped into the era of multi-core, providing a good chance to spread the parallel discrete event simulation. The parallel programming model and synchronization problem during the parallelization of discrete event simulation on multi-core platform were discussed. A parallel discrete event simulator based on multi-core platform was designed and implemented using the optimistic synchronization algorithm. On an HP multi-core server with up to 8 cores, both the overheads of the parallel simulator and the effects of event granularity, process number, lookahead on the simulation performance were tested using the Phold model. The experiment results show that the optimistic parallel discrete event simulation based on multi-core platform could achieve good speedup for simulation applications with coarse-grained events.