Adversarial examples present significant challenges to the security of Deep Neural Network (DNN) applications. Specifically, there are patch-based and texture-based attacks that are usually used to craft physical-world adversarial examples, posing real threats to security-critical applications such as person detection in surveillance and autonomous systems, because those attacks are physically realizable. Existing defense mechanisms face challenges in the adaptive attack setting, i.e., the attacks are specifically designed against them. In this paper, we propose Adversarial Spectrum Defense (ASD), a defense mechanism that leverages spectral decomposition via Discrete Wavelet Transform (DWT) to analyze adversarial patterns across multiple frequency scales. The multi-resolution and localization capability of DWT enables ASD to capture both high-frequency (fine-grained) and low-frequency (spatially pervasive) perturbations. By integrating this spectral analysis with the off-the-shelf Adversarial Training (AT) model, ASD provides a comprehensive defense strategy against both patch-based and texture-based adversarial attacks. Extensive experiments demonstrate that ASD+AT achieved state-of-the-art (SOTA) performance against various attacks, outperforming the APs of previous defense methods by 21.73%, in the face of strong adaptive adversaries specifically designed against ASD.

Intelligent Transportation Systems (ITS) utilises the growing trend of both communication technologies and intelligent analytics to make transportation systems more smart and efficient. Federated Learning, a privacy-preserving machine learning paradigm shows promise in being applied in this field. However, the data and device heterogeneity, and highly dynamic environment in ITS pose challenges to the performance of federated learning. One of the recent approaches to address the challenges are to choose proper participants from available clients during training. However, this research field is not fully investigated yet, and many works are still based on the classic random-based selection scheme. In this paper, we present Newt, an enhanced federated learning approach. On one hand, it includes a new client selection utility that explores the trade-off between accuracy performance in each round and system progress. On the other hand, it highlights a feedback control on the selector. Specifically, we implement a control on the selection frequency as a new dimension of client selection method design. We evaluate the proposed system with DNN training tasks on large scale FEMNIST-based datasets that are of different heterogeneity properties. The experiments show that our method outperforms the other baseline methods by as large as 20%.

With the current large scale deployment of machine learning technologies, such as those on cloud servers and edge and IoT hardwares, machine learning systems have been widely prevalence. Practical requirement has driven their performance increase in both academia and industry. However, the application requirement varies greatly across different applications, and directly using off-the-shelf systems might not be sufficient in many cases. In this work, we first propose to implement a series of techniques to optimize performance of convolution operation, one of the most important operations, in constructing deep learning networks. Besides, we also propose to apply the automated empirical optimisation of software approach to improve the performance of operators in machine learning system, most notably across various hardware platforms. Evaluation compared to existing libraries on different hardware devices has proved the efficiency of our proposed method.

With the advancement of portable head-mounted displays, interest in educational application of immersive spherical video-based virtual reality (SVVR) has been emerging. However, it remains unclear regarding the effects of immersive SVVR on cognitive and affective outcomes. In this study, we retrieved 58 learning outcomes from 16 studies. A meta-analysis was performed using the random effects model to calculate the effect size. Several important moderators were also examined such as control group treatment, learning outcome type, interaction functionality, content instruction, learning domain, and learner's stage. The results show that immersive SVVR is more effective than other instructional conditions with a medium effect size. The key findings of the moderator analysis are that immersive SVVR has a greater impact on affective outcomes, as well as under the conditions that learning system provides interaction functionality or integrates with content instruction before virtual exploratory learning.

The wireless local area network indoor localisation method based on fingerprints has been widely researched and applied due to its higher positioning accuracy and lower cost. However, its engineering is limited because of the large offline workload and time‐varying signal. To address these problems, an indoor localisation algorithm based on Markov state iterative analysis (MSIA) and fingerprint clustering structural optimisation (FCSO) is proposed in this study. First, the received signal strength time variation can be solved using the MSIA algorithm, which is based on the correlation of the memory source. Then, the offline workload can be considerably reduced and positioning accuracy can be improved with the FCSO algorithm, which contains fingerprint structural and clustering optimisation stages. In the structural optimisation stage, about half of the fingerprints can be omitted. In the clustering optimisation stage, high errors can be avoided. Finally, the locations of the positioning point can be obtained through the combination of MSIA and FCSO. Experimental results show the proposed algorithm can reduce the offline workload by about 50%, and the positioning accuracy can be increased when using MSIA and FCSO algorithms compared with other algorithms.