Towards trustworthy multi-modal motion prediction:Holistic evaluation and interpretability of outputs

Predicting the motion of other road agents enables autonomous vehicles to perform safe and efficient path planning.This task is very complex,as the behaviour of road agents depends on many factors and the number of possible future trajectories can be consid-erable(multi-modal).Most prior approaches proposed to address multi-modal motion prediction are based on complex machine learning systems that have limited interpret-ability.Moreover,the metrics used in current benchmarks do not evaluate all aspects of the problem,such as the diversity and admissibility of the output.The authors aim to advance towards the design of trustworthy motion prediction systems,based on some of the re-quirements for the design of Trustworthy Artificial Intelligence.The focus is on evaluation criteria,robustness,and interpretability of outputs.First,the evaluation metrics are comprehensively analysed,the main gaps of current benchmarks are identified,and a new holistic evaluation framework is proposed.Then,a method for the assessment of spatial and temporal robustness is introduced by simulating noise in the perception system.To enhance the interpretability of the outputs and generate more balanced results in the proposed evaluation framework,an intent prediction layer that can be attached to multi-modal motion prediction models is proposed.The effectiveness of this approach is assessed through a survey that explores different elements in the visualisation of the multi-modal trajectories and intentions.The proposed approach and findings make a significant contribution to the development of trustworthy motion prediction systems for autono-mous vehicles,advancing the field towards greater safety and reliability.

CrossLinkNet: An Explainable and Trustworthy AI Framework for Whole-Slide Images Segmentation

In the intelligent medical diagnosis area,Artificial Intelligence(AI)’s trustworthiness,reliability,and interpretability are critical,especially in cancer diagnosis.Traditional neural networks,while excellent at processing natural images,often lack interpretability and adaptability when processing high-resolution digital pathological images.This limitation is particularly evident in pathological diagnosis,which is the gold standard of cancer diagnosis and relies on a pathologist’s careful examination and analysis of digital pathological slides to identify the features and progression of the disease.Therefore,the integration of interpretable AI into smart medical diagnosis is not only an inevitable technological trend but also a key to improving diagnostic accuracy and reliability.In this paper,we introduce an innovative Multi-Scale Multi-Branch Feature Encoder(MSBE)and present the design of the CrossLinkNet Framework.The MSBE enhances the network’s capability for feature extraction by allowing the adjustment of hyperparameters to configure the number of branches and modules.The CrossLinkNet Framework,serving as a versatile image segmentation network architecture,employs cross-layer encoder-decoder connections for multi-level feature fusion,thereby enhancing feature integration and segmentation accuracy.Comprehensive quantitative and qualitative experiments on two datasets demonstrate that CrossLinkNet,equipped with the MSBE encoder,not only achieves accurate segmentation results but is also adaptable to various tumor segmentation tasks and scenarios by replacing different feature encoders.Crucially,CrossLinkNet emphasizes the interpretability of the AI model,a crucial aspect for medical professionals,providing an in-depth understanding of the model’s decisions and thereby enhancing trust and reliability in AI-assisted diagnostics.

Uncertainty-Aware Deep Learning: A Promising Tool for Trustworthy Fault Diagnosis

Recently,intelligent fault diagnosis based on deep learning has been extensively investigated,exhibiting state-of-the-art performance.However,the deep learning model is often not truly trusted by users due to the lack of interpretability of“black box”,which limits its deployment in safety-critical applications.A trusted fault diagnosis system requires that the faults can be accurately diagnosed in most cases,and the human in the deci-sion-making loop can be found to deal with the abnormal situa-tion when the models fail.In this paper,we explore a simplified method for quantifying both aleatoric and epistemic uncertainty in deterministic networks,called SAEU.In SAEU,Multivariate Gaussian distribution is employed in the deep architecture to compensate for the shortcomings of complexity and applicability of Bayesian neural networks.Based on the SAEU,we propose a unified uncertainty-aware deep learning framework(UU-DLF)to realize the grand vision of trustworthy fault diagnosis.Moreover,our UU-DLF effectively embodies the idea of“humans in the loop”,which not only allows for manual intervention in abnor-mal situations of diagnostic models,but also makes correspond-ing improvements on existing models based on traceability analy-sis.Finally,two experiments conducted on the gearbox and aero-engine bevel gears are used to demonstrate the effectiveness of UU-DLF and explore the effective reasons behind.

Adversarial Attack-Based Robustness Evaluation for Trustworthy AI

Artificial Intelligence(AI)technology has been extensively researched in various fields,including the field of malware detection.AI models must be trustworthy to introduce AI systems into critical decisionmaking and resource protection roles.The problem of robustness to adversarial attacks is a significant barrier to trustworthy AI.Although various adversarial attack and defense methods are actively being studied,there is a lack of research on robustness evaluation metrics that serve as standards for determining whether AI models are safe and reliable against adversarial attacks.An AI model’s robustness level cannot be evaluated by traditional evaluation indicators such as accuracy and recall.Additional evaluation indicators are necessary to evaluate the robustness of AI models against adversarial attacks.In this paper,a Sophisticated Adversarial Robustness Score(SARS)is proposed for AI model robustness evaluation.SARS uses various factors in addition to the ratio of perturbated features and the size of perturbation to evaluate robustness accurately in the evaluation process.This evaluation indicator reflects aspects that are difficult to evaluate using traditional evaluation indicators.Moreover,the level of robustness can be evaluated by considering the difficulty of generating adversarial samples through adversarial attacks.This paper proposed using SARS,calculated based on adversarial attacks,to identify data groups with robustness vulnerability and improve robustness through adversarial training.Through SARS,it is possible to evaluate the level of robustness,which can help developers identify areas for improvement.To validate the proposed method,experiments were conducted using a malware dataset.Through adversarial training,it was confirmed that SARS increased by 70.59%,and the recall reduction rate improved by 64.96%.Through SARS,it is possible to evaluate whether an AI model is vulnerable to adversarial attacks and to identify vulnerable data types.In addition,it is expected that improved models can be achieved by improving resistance to adversarial attacks via methods such as adversarial training.

Trustworthy Metrics Models for Internetware

To measure the trustworthiness of Intemetware, we need to understand the existing problems and design appropriate trustworthy metrics. The developing and running system of Internetware is analyzed in terms of process, keystone, methods and techniques. According to the main related factors of Internetware trustworthiness, two important models, namely trustworthy metrics hierarchy model of components （TMHMC） with computing steps and local-global trustworthy metrics model of platform （LGTMMP） with algorithm respectively, are employed to evaluate the internal and external trustworthiness of Internetware benefiting for the development of Internetware.

SATVPC:Secure-agent-based trustworthy virtual private cloud model in open computing environments

Private clouds and public clouds are turning mutually into the open integrated cloud computing environment,which can aggregate and utilize WAN and LAN networks computing,storage,information and other hardware and software resources sufficiently,but also bring a series of security,reliability and credibility problems.To solve these problems,a novel secure-agent-based trustworthy virtual private cloud model named SATVPC was proposed for the integrated and open cloud computing environment.Through the introduction of secure-agent technology,SATVPC provides an independent,safe and trustworthy computing virtual private platform for multi-tenant systems.In order to meet the needs of the credibility of SATVPC and mandate the trust relationship between each task execution agent and task executor node suitable for their security policies,a new dynamic composite credibility evaluation mechanism was presented,including the credit index computing algorithm and the credibility differentiation strategy.The experimental system shows that SATVPC and the credibility evaluation mechanism can ensure the security of open computing environments with feasibility.Experimental results and performance analysis also show that the credit indexes computing algorithm can evaluate the credibilities of task execution agents and task executor nodes quantitatively,correctly and operationally.

User preferences-aware recommendation for trustworthy cloud services based on fuzzy clustering

The cloud computing has been growing over the past few years, and service providers are creating an intense competitive world of business. This proliferation makes it hard for new users to select a proper service among a large amount of service candidates. A novel user preferences-aware recommendation approach for trustworthy services is presented. For describing the requirements of new users in different application scenarios, user preferences are identified by usage preference, trust preference and cost preference. According to the similarity analysis of usage preference between consumers and new users, the candidates are selected, and these data about service trust provided by them are calculated as the fuzzy comprehensive evaluations. In accordance with the trust and cost preferences of new users, the dynamic fuzzy clusters are generated based on the fuzzy similarity computation. Then, the most suitable services can be selected to recommend to new users. The experiments show that this approach is effective and feasible, and can improve the quality of services recommendation meeting the requirements of new users in different scenario.

CHC-TSCM:A Trustworthy Service Composition Method Based on an Improved CHC Genetic Algorithm

Trustworthy service composition is an extremely important task when service composition becomes infeasible or even fails in an environment which is open,autonomic,uncertain and deceptive.This paper presents a trustworthy service composition method based on an improved Cross generation elitist selection,Heterogeneous recombination,Catacly-smic mutation(CHC) Trustworthy Service Composition Method(CHC-TSCM) genetic algorithm.CHCTSCM firstly obtains the total trust degree of the individual service using a trust degree measurement and evaluation model proposed in previous research.Trust combination and computation then are performed according to the structural relation of the composite service.Finally,the optimal trustworthy service composition is acquired by the improved CHC genetic algorithm.Experimental results show that CHC-TSCM can effectively solve the trustworthy service composition problem.Comparing with GODSS and TOCSS,this new method has several advantages:1) a higher service composition successrate;2) a smaller decline trend of the service composition success-rate,and 3) enhanced stability.

Security,Controllability,Manageability and Survivability in Trustworthy Network

The Internet plays increasingly important roles in everyone's life; however, the existence of a mismatch between the basic architectural idea beneath the Internet and the emerging requirements for it is becoming more and more obvious. Although the Internet community came up with a consensus that the future network should be trustworthy, the concept of 'trustworthy networks' and the ways leading us to a trustworthy network are not yet clear. This research insists that the security, controllability, manageability, and survivability should be basic properties of a trustworthy network. The key ideas and techniques involved in these properties are studied, and recent developments and progresses are surveyed. At the same time, the technical trends and challenges are briefly discussed. The network trustworthiness could and should be eventually achieved.

A personalized trustworthy service selection method

A personalized trustworthy service selection method is proposed to fully express the features of trust, emphasize the importance of user preference and improve the trustworthiness of service selection. The trustworthiness of web service is defined as customized multi-dimensional trust metrics and the user preference is embodied in the weight of each trust metric. A service selection method combining AHP （analytic hierarchy process） and PROMETHEE （preference ranking organization method for enrichment evaluations） is proposed. AHP is used to determine the weights of trust metrics according to users＇ preferences. Hierarchy and pairwise comparison matrices are constructed. The weights of trust metrics are derived from the highest eigenvalue and eigenvector of the matrix. to obtain the final rank of candidate services. The preference functions are defined according to the inherent characteristics of the trust metrics and net outranking flows are calculated. Experimental results show that the proposed method can effectively express users＇ personalized preferences for trust metrics, and the trustworthiness of service ranking and selection is efficiently improved.

Quality of Chinese Export Toys Is Trustworthy

On September 4th, 2007, AQSIQ and the Press Office of the State Council invited 13 media representatives from countries such as America, Britain, France, Japan, Canada, Singapore to visit the Technical Center Toy Laboratory of Guangdong Entry-exit Inspection and Quarantine Bureau, Zhentai (China) Industrial Limited Company and Guangdong Xinboxing Toys Limited Company.……

Trustworthy Artificial Intelligence for Social Governance

As AI technology continues to evolve,it plays an increasingly significant role in everyday life and social governance.However,the frequent occurrence of issues such as algorithmic bias,privacy breaches,and data leaks has led to a crisis of trust in AI among the public,presenting numerous challenges to social governance.Establishing technical trust in Al,reducing uncertainties in AI development,and enhancing its effectiveness in social governance have become a consensus among policymakers and researchers.By comparing different types of AI,the paper proposes and conceptualizes the idea of trustworthy Al,then discusses its characteristics and its value and impact pathways in social governance.The analysis addresses how mismatches in technological trust can affect social stability and the advancement of AI strategies.The paper highlights the potential of trustworthy AI to improve the efficiency of social governance and solve complex social problems.

Provenance Documentation to Enable Explainable and Trustworthy AI:A Literature Review

Recently artificial intelligence(AI)and machine learning(ML)models have demonstrated remarkable progress with applications developed in various domains.It is also increasingly discussed that AI and ML models and applications should be transparent,explainable,and trustworthy.Accordingly,the field of Explainable AI(XAI)is expanding rapidly.XAI holds substantial promise for improving trust and transparency in AI-based systems by explaining how complex models such as the deep neural network(DNN)produces their outcomes.Moreover,many researchers and practitioners consider that using provenance to explain these complex models will help improve transparency in AI-based systems.In this paper,we conduct a systematic literature review of provenance,XAI,and trustworthy AI(TAI)to explain the fundamental concepts and illustrate the potential of using provenance as a medium to help accomplish explainability in AI-based systems.Moreover,we also discuss the patterns of recent developments in this area and offer a vision for research in the near future.We hope this literature review will serve as a starting point for scholars and practitioners interested in learning about essential components of provenance,XAI,and TAI.

Trustworthy decentralized collaborative learning for edge intelligence:A survey

Edge intelligence is an emerging technology that enables artificial intelligence on connected systems and devices in close proximity to the data sources.decentralized collaborative learning(DCL)is a novel edge intelligence technique that allows distributed clients to cooperatively train a global learning model without revealing their data.DCL has a wide range of applications in various domains,such as smart city and autonomous driving.However,DCL faces significant challenges in ensuring its trustworthiness,as data isolation and privacy issues make DCL systems vulnerable to adversarial attacks that aim to breach system confidentiality,undermine learning reliability or violate data privacy.Therefore,it is crucial to design DCL in a trustworthy manner,with a focus on security,robustness,and privacy.In this survey,we present a comprehensive review of existing efforts for designing trustworthy DCL systems from the three key aformentioned aspects:security,robustness,and privacy.We analyze the threats that affect the trustworthiness of DCL across different scenarios and assess specific technical solutions for achieving each aspect of trustworthy DCL(TDCL).Finally,we highlight open challenges and future directions for advancing TDCL research and practice.

An Approach of Trustworthy Software Design with Automatically Adapting Software Update

In this paper, we merge software trustworthiness with software design and present an approach to trustworthy software design with an automatically adapting software update. First, software behavior and results can be expected and behavior states can be monitored when a software runs by introducing a trustworthy behavior trace on a software and inserting a checkpoint sensor at each checkpoint of the trustworthy software. Second, an updated approach of the trustworthy behavior trace for the software at the level of checkpoints is presented. The trustworthy behavior traces of the software for two versions can be merged adequately by constructing split points and merge points between two trustworthy behavior traces. Finally, experiments and analyses show that： （1） the software designed by our approach can detect and report the anomaly in a software automatically and effectively, so it has a higher ability of trustworthiness evaluation than the traditional software; and （2） our approach can realize the accurate update of the trustworthy behavior trace with a lower space overhead of checkpoints when the software updates.

FinB rain 2.0:when finance meets trustworthy AI

Artificial intelligence(AI) has accelerated the advancement of financial services by identifying hidden patterns from data to improve the quality of financial decisions. However, in addition to commonly desired attributes,such as model accuracy, financial services demand trustworthy AI with properties that have not been adequately realized. These properties of trustworthy AI are interpretability, fairness and inclusiveness, robustness and security,and privacy protection. Here, we review the recent progress and limitations of applying AI to various areas of financial services, including risk management, fraud detection, wealth management, personalized services, and regulatory technology. Based on these progress and limitations, we introduce FinBrain 2.0, a research framework toward trustworthy AI. We argue that we are still a long way from having a truly trustworthy AI in financial services and call for the communities of AI and financial industry to join in this effort.

Trustworthy Mobile Network Architecture Based on Locator/Identity Split

Today＇s Internet architecture faces many challenges, from poor support for mobility to security threats. By analyzing the drawbacks of current TCP/IP protocol, we propose a new network architecture model LISNA. LISNA defines a kind of network architecture with mobility and trustworthiness, based upon the decoupling of end-host identity information from location information. This paper makes a brief description of the basic concepts and model structure that support network mobility and security. After introducing the key techniques in LISNA, the paper further illustrates how to promote the mobility handoff and to keep security association.

Recommend trustworthy services using interval numbers of four parameters via cloud model for potential users

How to discover the trustworthy services is a challenge for potential users because of the deficiency of us- age experiences and the information overload of QoE （qual- ity of experience） evaluations from consumers. Aiming to the limitations of traditional interval numbers in measuring the trustworthiness of service, this paper proposed a novel ser- vice recommendation approach using the interval numbers of four parameters （INF） for potential users. In this approach, a trustworthiness cloud model was established to identify the eigenvalue of INF via backward cloud generator, and a new formula of INF possibility degree based on geometrical anal- ysis was presented to ensure the high calculation precision. In order to select the highly valuable QoE evaluations, the similarity of client-side feature between potential user and consumers was calculated, and the multi-attributes trustwor- thiness values were aggregated into INF by the fuzzy ana- lytic hierarchy process method. On the basis of ranking INF, the sort values of trustworthiness of candidate services were obtained, and the trustworthy services were chosen to recommend to potential user. The experiments based on a real-world dataset showed that it can improve the recommendation accuracy of trustworthy services compared to other approaches, which contributes to solving cold start and information overload problem in service recommendation.

Research and Analysis of Structural Hole and Matching Coefficient

Measure is a map from the reality or experimental world to the mathematical world, through which people can more easily understand the properties of entities and the relationship between them. But the traditional software measurement methods have been unable to effectively measure this large-scale software. Therefore, trustworthy measurement gives an accurate measurement to these emerging features, providing valuable perspectives and different research dimensions to understand software systems. The paper introduces the complex network theory to software measurement methods and proposes a statistical measurement methodology. First we study the basic parameters of the complex network, and then introduce two new measurement parameters: structural holes, matching coefficient.

Study on the Current Consumption Environment and Its Improvement Strategies in China

This paper analyses the current consumption environment in China, gives a full picture of the protection of Chinese consumers＇ rights and interests, especially describes some problems that damage consumers＇ legitimate rights and interests. This paper also puts forward some practical meanings of improvement of the consumption environment in order to make it more and more trustworthy in China, and then seek to provide some useful suggestions and approaches constructively from different parts to this campaign.