Deep learning techniques were widely adopted in various scenarios as a service. However, they are found naturally exposed to adversarial attacks. Such imperceptible-perturbation-based attacks can cause severe damage in nowaday authentication systems that adopt DNNs as the core, such as fingerprint liveness detection systems, face recognition systems, etc. This paper avoids improving the model's robustness and realizes the defense against adversarial attacks based on denoising and reconstruction. Our proposed method can be viewed as a two-step defense framework. The first step denoises the input adversarial example, then reconstructing the sample to close to the original clean image and help the target model output the original label. The proposed method is evaluated using six kinds of state-of-art adversarial attacks, including the adaptive attacks, which are known as the strongest attacks.We also specifically focus on demonstrating the effectiveness of our proposed work in Finance Authentication systems as a real-life case study. Experimental results reveal that our method is more robust than the previous super-resolution-only defense in respect of attaining a higher averaging accuracy over clean and distorted samples. To the best of our knowledge, it's the first work that reveals a comprehensive defense framework against adversarial attacks over Finance Authentication systems.

Teleconsultation is conducive to the rational use of health resources and plays a key role in the realization of hierarchical diagnosis and treatment. Previous research on telemedicine multicriteria decision making methods rarely consider the difference in responsibilities between remote experts and attending physicians. To show the difference precisely, this article first defines the intuitionistic hesitant fuzzy linguistic term set (IHFLTS) as an extended mathematical expression of the fuzzy set. We propose two kinds of correlation measuring methods and related hesitancy degrees based on information energy and statistics with different application scenarios. Based on information entropy and distance measurement, two criterion weight determination methods are studied. To formally implement the joint decision-making process of remote experts and attending physicians, a telemedicine decision-making model is established for designing the teleconsultation decision support system. We apply the proposed methods to a sinusitis teleconsultation case to demonstrate their effectiveness. The weight determination methods are applied, and the influence of parameter variation in the distance-based method is analyzed. Finally, the rationality and applicability of the proposed methods are verified by comparisons. The results show that by considering the importance and differences between the two roles of teleconsultation, IHFLTS can obtain a more realistic result for the referral problems.