超过人类决策能力的机器学习模型的出现，在复杂的领域中启动了一种运动，以构建与人类互动的AI系统。许多构建基础对于这项活动至关重要，中心是人类行为的算法表征。尽管现有的大部分工作都集中在人类的总体行为上，但一个重要的远程目标是开发专门针对个人人并可以在其中区分的行为模型。为了使这个过程形式化，我们研究了行为风格的问题，其中任务是仅从决策中确定决策者。我们提出了一种基于变压器的方法，用于在国际象棋的背景下进行行为风格测量法，其中有人试图识别玩一组游戏的玩家。我们的方法在几个弹药的分类框架中运行，并且可以在只有100个标签游戏的情况下正确地从成千上万的候选玩家中识别出98％精度的候选人。即使接受业余比赛的训练，我们的方法还是对大师级玩家的分布样本的概括，尽管业余球员和世界一流的球员之间存在巨大差异。最后，我们更广泛地考虑了我们所产生的嵌入有关国际象棋中人类风格的揭示的内容，以及在行为数据中识别个人的强大方法的潜在伦理含义。

人工智能研究中的一个新兴主题是创建模型，以模拟特定人员的决策和行为，包括游戏玩法，文本生成和艺术表达。这些模型以对个人的量身定制的方式以及为互动而不是简单地繁殖固定的预计行为的复制方式而超越了早期的方法。我们将这些称为模拟模型，在本文中，我们开发了一个框架，以表征其日益增长的可用性所带来的道德和社会问题。我们的框架包括用于使用此类模型的许多不同方案，并考虑了对一系列不同参与者的影响，包括正在建模的目标，部署模型的操作员以及与之交互的实体。

在许多实际情况下，数据分布在多个位置，不能组合培训。联合学习是一种新型的分布式学习方法，允许多个联合代理共同学习模型。虽然这种方法可能会减少每个代理经历的错误，但它也提出了公平的问题：一个代理经历的错误在多大程度上可以明显低于另一代理经历的错误？在这项工作中，我们考虑了两个公平的概念，每个公平可能在不同的情况下适当：“平等主义公平”（旨在束缚误差率如何）和“比例公平”（这旨在奖励参与者促进更多数据）。对于平等主义公平性，我们获得了一个紧张的乘法界，就如何在联合联盟的代理之间发散了差异。对于比例公平性，我们表明，对于任何单独的合理联合联盟，我们表明次级比例误差（相对于有贡献的数据点数）被保证。

在人类可能希望从这些系统中学习，与它们合作或作为合作伙伴互动的情况下，可以捕获类似人类行为的AI系统越来越有用。为了开发以人为导向的AI系统，预测人类行为（而不是预测最佳行动）的问题受到了广泛关注。现有的工作集中在总体意义上捕获人类行为，这可能会限制任何特定个人可以从与这些系统互动中获得的收益。我们通过开发国际象棋中人类行为的高度准确的预测模型来扩展这一工作。国际象棋是探索人类互动的一个丰富领域，因为它结合了一套独特的属性：AI系统在多年前实现了超人类的表现，但人类仍然与他们以及对手和准备工具紧密互动，并且有一种关于单个玩家游戏的大量记录数据。从迈亚（Maia）开始，该版本的Alphazero经过了对人类人群的培训，我们证明我们可以通过应用一系列微调方法来显着提高特定玩家的举动的预测准确性。此外，我们的个性化模型可用于执行风格测定法 - 预测谁采取了一组给定的动作 - 表明他们在个人层面上捕获了人类的决策。我们的工作展示了一种使AI系统更好地与个人行为保持一致的方法，这可能会导致人类互动的大量改善。

Transfer learning from natural image datasets, particularly I N , using standard large models and corresponding pretrained weights has become a de-facto method for deep learning applications to medical imaging. However, there are fundamental di erences in data sizes, features and task speci cations between natural image classi cation and the target medical tasks, and there is little understanding of the e ects of transfer. In this paper, we explore properties of transfer learning for medical imaging. A performance evaluation on two large scale medical imaging tasks shows that surprisingly, transfer o ers little bene t to performance, and simple, lightweight models can perform comparably to I N architectures. Investigating the learned representations and features, we nd that some of the di erences from transfer learning are due to the over-parametrization of standard models rather than sophisticated feature reuse. We isolate where useful feature reuse occurs, and outline the implications for more e cient model exploration. We also explore feature independent bene ts of transfer arising from weight scalings.

The machine learning community has become increasingly concerned with the potential for bias and discrimination in predictive models. This has motivated a growing line of work on what it means for a classification procedure to be "fair." In this paper, we investigate the tension between minimizing error disparity across different population groups while maintaining calibrated probability estimates. We show that calibration is compatible only with a single error constraint (i.e. equal false-negatives rates across groups), and show that any algorithm that satisfies this relaxation is no better than randomizing a percentage of predictions for an existing classifier. These unsettling findings, which extend and generalize existing results, are empirically confirmed on several datasets. * Equal contribution, alphebetical order. 1 For the remainder of the paper, we will use Equalized Odds to refer to this notion of non-discrimination.

Recent discussion in the public sphere about algorithmic classification has involved tension between competing notions of what it means for a probabilistic classification to be fair to different groups. We formalize three fairness conditions that lie at the heart of these debates, and we prove that except in highly constrained special cases, there is no method that can satisfy these three conditions simultaneously. Moreover, even satisfying all three conditions approximately requires that the data lie in an approximate version of one of the constrained special cases identified by our theorem. These results suggest some of the ways in which key notions of fairness are incompatible with each other, and hence provide a framework for thinking about the trade-offs between them.

This paper expounds the design and control of a new Variable Stiffness Series Elastic Actuator (VSSEA). It is established by employing a modular mechanical design approach that allows us to effectively optimise the stiffness modulation characteristics and power density of the actuator. The proposed VSSEA possesses the following features: i) no limitation in the work-range of output link, ii) a wide range of stiffness modulation (~20Nm/rad to ~1KNm/rad), iii) low-energy-cost stiffness modulation at equilibrium and non-equilibrium positions, iv) compact design and high torque density (~36Nm/kg), and v) high-speed stiffness modulation (~3000Nm/rad/s). Such features can help boost the safety and performance of many advanced robotic systems, e.g., a cobot that physically interacts with unstructured environments and an exoskeleton that provides physical assistance to human users. These features can also enable us to utilise variable stiffness property to attain various regulation and trajectory tracking control tasks only by employing conventional controllers, eliminating the need for synthesising complex motion control systems in compliant actuation. To this end, it is experimentally demonstrated that the proposed VSSEA is capable of precisely tracking desired position and force control references through the use of conventional Proportional-Integral-Derivative (PID) controllers.

Targeted syntactic evaluations of language models ask whether models show stable preferences for syntactically acceptable content over minimal-pair unacceptable inputs. Most targeted syntactic evaluation datasets ask models to make these judgements with just a single context-free sentence as input. This does not match language models' training regime, in which input sentences are always highly contextualized by the surrounding corpus. This mismatch raises an important question: how robust are models' syntactic judgements in different contexts? In this paper, we investigate the stability of language models' performance on targeted syntactic evaluations as we vary properties of the input context: the length of the context, the types of syntactic phenomena it contains, and whether or not there are violations of grammaticality. We find that model judgements are generally robust when placed in randomly sampled linguistic contexts. However, they are substantially unstable for contexts containing syntactic structures matching those in the critical test content. Among all tested models (GPT-2 and five variants of OPT), we significantly improve models' judgements by providing contexts with matching syntactic structures, and conversely significantly worsen them using unacceptable contexts with matching but violated syntactic structures. This effect is amplified by the length of the context, except for unrelated inputs. We show that these changes in model performance are not explainable by simple features matching the context and the test inputs, such as lexical overlap and dependency overlap. This sensitivity to highly specific syntactic features of the context can only be explained by the models' implicit in-context learning abilities.

Deepfakes are computationally-created entities that falsely represent reality. They can take image, video, and audio modalities, and pose a threat to many areas of systems and societies, comprising a topic of interest to various aspects of cybersecurity and cybersafety. In 2020 a workshop consulting AI experts from academia, policing, government, the private sector, and state security agencies ranked deepfakes as the most serious AI threat. These experts noted that since fake material can propagate through many uncontrolled routes, changes in citizen behaviour may be the only effective defence. This study aims to assess human ability to identify image deepfakes of human faces (StyleGAN2:FFHQ) from nondeepfake images (FFHQ), and to assess the effectiveness of simple interventions intended to improve detection accuracy. Using an online survey, 280 participants were randomly allocated to one of four groups: a control group, and 3 assistance interventions. Each participant was shown a sequence of 20 images randomly selected from a pool of 50 deepfake and 50 real images of human faces. Participants were asked if each image was AI-generated or not, to report their confidence, and to describe the reasoning behind each response. Overall detection accuracy was only just above chance and none of the interventions significantly improved this. Participants' confidence in their answers was high and unrelated to accuracy. Assessing the results on a per-image basis reveals participants consistently found certain images harder to label correctly, but reported similarly high confidence regardless of the image. Thus, although participant accuracy was 62% overall, this accuracy across images ranged quite evenly between 85% and 30%, with an accuracy of below 50% for one in every five images. We interpret the findings as suggesting that there is a need for an urgent call to action to address this threat.