There is general agreement that some form of regulation is necessary both for AI creators to be incentivised to develop trustworthy systems, and for users to actually trust those systems. But there is much debate about what form these regulations should take and how they should be implemented. Most work in this area has been qualitative, and has not been able to make formal predictions. Here, we propose that evolutionary game theory can be used to quantitatively model the dilemmas faced by users, AI creators, and regulators, and provide insights into the possible effects of different regulatory regimes. We show that creating trustworthy AI and user trust requires regulators to be incentivised to regulate effectively. We demonstrate the effectiveness of two mechanisms that can achieve this. The first is where governments can recognise and reward regulators that do a good job. In that case, if the AI system is not too risky for users then some level of trustworthy development and user trust evolves. We then consider an alternative solution, where users can condition their trust decision on the effectiveness of the regulators. This leads to effective regulation, and consequently the development of trustworthy AI and user trust, provided that the cost of implementing regulations is not too high. Our findings highlight the importance of considering the effect of different regulatory regimes from an evolutionary game theoretic perspective.
相關內容
We propose an interdisciplinary framework that combines Bayesian predictive inference, a well-established tool in Machine Learning, with Formal Methods rooted in the computer science community. Bayesian predictive inference allows for coherently incorporating uncertainty about unknown quantities by making use of methods or models that produce predictive distributions, which in turn inform decision problems. By formalizing these decision problems into properties with the help of spatio-temporal logic, we can formulate and predict how likely such properties are to be satisfied in the future at a certain location. Moreover, we can leverage our methodology to evaluate and compare models directly on their ability to predict the satisfaction of application-driven properties. The approach is illustrated in an urban mobility application, where the crowdedness in the center of Milan is proxied by aggregated mobile phone traffic data. We specify several desirable spatio-temporal properties related to city crowdedness such as a fault-tolerant network or the reachability of hospitals. After verifying these properties on draws from the posterior predictive distributions, we compare several spatio-temporal Bayesian models based on their overall and property-based predictive performance.
We consider nonparametric statistical inference on a periodic interaction potential $W$ from noisy discrete space-time measurements of solutions $\rho=\rho_W$ of the nonlinear McKean-Vlasov equation, describing the probability density of the mean field limit of an interacting particle system. We show how Gaussian process priors assigned to $W$ give rise to posterior mean estimators that exhibit fast convergence rates for the implied estimated densities $\bar \rho$ towards $\rho_W$. We further show that if the initial condition $\phi$ is not too smooth and satisfies a standard deconvolvability condition, then one can consistently infer the potential $W$ itself at convergence rates $N^{-\theta}$ for appropriate $\theta>0$, where $N$ is the number of measurements. The exponent $\theta$ can be taken to approach $1/2$ as the regularity of $W$ increases corresponding to `near-parametric' models.
The collaboration in co-located shared environments has sparked an increased interest in immersive technologies, including Augmented Reality (AR). Since research in this field has primarily focused on individual user experiences in AR, the collaborative aspects within shared AR spaces remain less explored, and fewer studies can provide guidelines for designing this type of experience. This article investigates how the user experience in a collaborative shared AR space is affected by divergent perceptions of virtual objects and the effects of positional synchrony and avatars. For this purpose, we developed an AR app and used two distinct experimental conditions to study the influencing factors. Forty-eight participants, organized into 24 pairs, participated in the experiment and jointly interacted with shared virtual objects. Results indicate that divergent perceptions of virtual objects did not directly influence communication and collaboration dynamics. Conversely, positional synchrony emerged as a critical factor, significantly enhancing the quality of the collaborative experience. On the contrary, while not negligible, avatars played a relatively less pronounced role in influencing these dynamics. The findings can potentially offer valuable practical insights, guiding the development of future collaborative AR/VR environments.
In safety-critical systems that interface with the real world, the role of uncertainty in decision-making is pivotal, particularly in the context of machine learning models. For the secure functioning of Cyber-Physical Systems (CPS), it is imperative to manage such uncertainty adeptly. In this research, we focus on the development of a vehicle's lateral control system using a machine learning framework. Specifically, we employ a Bayesian Neural Network (BNN), a probabilistic learning model, to address uncertainty quantification. This capability allows us to gauge the level of confidence or uncertainty in the model's predictions. The BNN based controller is trained using simulated data gathered from the vehicle traversing a single track and subsequently tested on various other tracks. We want to share two significant results: firstly, the trained model demonstrates the ability to adapt and effectively control the vehicle on multiple similar tracks. Secondly, the quantification of prediction confidence integrated into the controller serves as an early-warning system, signaling when the algorithm lacks confidence in its predictions and is therefore susceptible to failure. By establishing a confidence threshold, we can trigger manual intervention, ensuring that control is relinquished from the algorithm when it operates outside of safe parameters.
Auto-labeling is an important family of techniques that produce labeled training sets with minimum manual labeling. A prominent variant, threshold-based auto-labeling (TBAL), works by finding a threshold on a model's confidence scores above which it can accurately label unlabeled data points. However, many models are known to produce overconfident scores, leading to poor TBAL performance. While a natural idea is to apply off-the-shelf calibration methods to alleviate the overconfidence issue, such methods still fall short. Rather than experimenting with ad-hoc choices of confidence functions, we propose a framework for studying the \emph{optimal} TBAL confidence function. We develop a tractable version of the framework to obtain \texttt{Colander} (Confidence functions for Efficient and Reliable Auto-labeling), a new post-hoc method specifically designed to maximize performance in TBAL systems. We perform an extensive empirical evaluation of our method \texttt{Colander} and compare it against methods designed for calibration. \texttt{Colander} achieves up to 60\% improvements on coverage over the baselines while maintaining auto-labeling error below $5\%$ and using the same amount of labeled data as the baselines.
Technology presents a significant educational opportunity, particularly in enhancing emotional engagement and expanding learning and educational prospects for individuals with Neurodevelopmental Disorders (NDD). Virtual reality emerges as a promising tool for addressing such disorders, complemented by numerous touchscreen applications that have shown efficacy in fostering education and learning abilities. VR and touchscreen technologies represent diverse interface modalities. This study primarily investigates which interface, VR or touchscreen, more effectively facilitates food education for individuals with NDD. We compared learning outcomes via pre- and post-exposure questionnaires. To this end, we developed GEA, a dual-interface, user-friendly web application for Food Education, adaptable for either immersive use in a head-mounted display (HMD) or non-immersive use on a tablet. A controlled study was conducted to determine which interface better promotes learning. Over three sessions, the experimental group engaged with all GEA games in VR (condition A), while the control group interacted with the same games on a tablet (condition B). Results indicated a significant increase in post-questionnaire scores across subjects, averaging a 46% improvement. This enhancement was notably consistent between groups, with VR and Tablet groups showing 42% and 41% improvements, respectively.
In mesh simplification, common requirements like accuracy, triangle quality, and feature alignment are often considered as a trade-off. Existing algorithms concentrate on just one or a few specific aspects of these requirements. For example, the well-known Quadric Error Metrics (QEM) approach prioritizes accuracy and can preserve strong feature lines/points as well but falls short in ensuring high triangle quality and may degrade weak features that are not as distinctive as strong ones. In this paper, we propose a smooth functional that simultaneously considers all of these requirements. The functional comprises a normal anisotropy term and a Centroidal Voronoi Tessellation (CVT) energy term, with the variables being a set of movable points lying on the surface. The former inherits the spirit of QEM but operates in a continuous setting, while the latter encourages even point distribution, allowing various surface metrics. We further introduce a decaying weight to automatically balance the two terms. We selected 100 CAD models from the ABC dataset, along with 21 organic models, to compare the existing mesh simplification algorithms with ours. Experimental results reveal an important observation: the introduction of a decaying weight effectively reduces the conflict between the two terms and enables the alignment of weak features. This distinctive feature sets our approach apart from most existing mesh simplification methods and demonstrates significant potential in shape understanding.
Unsupervised person re-identification (Re-ID) attracts increasing attention due to its potential to resolve the scalability problem of supervised Re-ID models. Most existing unsupervised methods adopt an iterative clustering mechanism, where the network was trained based on pseudo labels generated by unsupervised clustering. However, clustering errors are inevitable. To generate high-quality pseudo-labels and mitigate the impact of clustering errors, we propose a novel clustering relationship modeling framework for unsupervised person Re-ID. Specifically, before clustering, the relation between unlabeled images is explored based on a graph correlation learning (GCL) module and the refined features are then used for clustering to generate high-quality pseudo-labels.Thus, GCL adaptively mines the relationship between samples in a mini-batch to reduce the impact of abnormal clustering when training. To train the network more effectively, we further propose a selective contrastive learning (SCL) method with a selective memory bank update policy. Extensive experiments demonstrate that our method shows much better results than most state-of-the-art unsupervised methods on Market1501, DukeMTMC-reID and MSMT17 datasets. We will release the code for model reproduction.
Although measuring held-out accuracy has been the primary approach to evaluate generalization, it often overestimates the performance of NLP models, while alternative approaches for evaluating models either focus on individual tasks or on specific behaviors. Inspired by principles of behavioral testing in software engineering, we introduce CheckList, a task-agnostic methodology for testing NLP models. CheckList includes a matrix of general linguistic capabilities and test types that facilitate comprehensive test ideation, as well as a software tool to generate a large and diverse number of test cases quickly. We illustrate the utility of CheckList with tests for three tasks, identifying critical failures in both commercial and state-of-art models. In a user study, a team responsible for a commercial sentiment analysis model found new and actionable bugs in an extensively tested model. In another user study, NLP practitioners with CheckList created twice as many tests, and found almost three times as many bugs as users without it.
Knowledge graphs capture structured information and relations between a set of entities or items. As such they represent an attractive source of information that could help improve recommender systems. However existing approaches in this domain rely on manual feature engineering and do not allow for end-to-end training. Here we propose knowledge-aware graph neural networks with label smoothness regularization to provide better recommendations. Conceptually, our approach computes user-specific item embeddings by first applying a trainable function that identifies important knowledge graph relationships for a given user. This way we transform the knowledge graph into a user-specific weighted graph and then applies a graph neural network to compute personalized item embeddings. To provide better inductive bias, we use label smoothness, which assumes that adjacent items in the knowledge graph are likely to have similar user relevance labels/scores. Label smoothness provides regularization over edge weights and we prove that it is equivalent to a label propagation scheme on a graph. Finally, we combine knowledge-aware graph neural networks and label smoothness and present the unified model. Experiment results show that our method outperforms strong baselines in four datasets. It also achieves strong performance in the scenario where user-item interactions are sparse.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191