A.1 Experimental design Figure 1 summarizes the experimental design used for our experiments. The participants that went through our experiments are users from the online platform Amazon Mechanical Turk (AMT). Through this platform, users stay anonymous, hence, we do not collect any sensitive personal information about them. We prioritized users with a Master qualification (which is a qualification attributed by AMT to users who have proven to be of excellent quality) or normal users with high qualifications (number of HIT completed = 10000and HIT accepted > 98%). Before going through the experiment, participants are asked to read and agree to a consent form, which specifies: the objective and procedure of the experiment, as well as the time expected to completion ( 5 - 8 min) with the reward associated ($1.4), and finally, the risk, benefits, and confidentiality of taking part in this study.

We propose a new perspective on representation learning in reinforcement learning based on geometric properties of the space of value functions. From there, we provide formal evidence regarding the usefulness of value functions as auxiliary tasks in reinforcement learning. Our formulation considers adapting the representation to minimize the (linear) approximation of the value function of all stationary policies for a given environment. We show that this optimization reduces to making accurate predictions regarding a special class of value functions which we call adversarial value functions (AVFs). We demonstrate that using value functions as auxiliary tasks corresponds to an expected-error relaxation of our formulation, with AVFs a natural candidate, and identify a close relationship with proto-value functions (Mahadevan, 2005). We highlight characteristics of AVFs and their usefulness as auxiliary tasks in a series of experiments on the four-room domain.

The existence of adversarial examples has been a mystery for years and attracted much interest. A well-known theory by \citet{ilyas2019adversarial} explains adversarial vulnerability from a data perspective by showing that one can extract non-robust features from adversarial examples and these features alone are useful for classification. However, the explanation remains quite counter-intuitive since non-robust features are mostly noise features to humans. In this paper, we re-examine the theory from a larger context by incorporating multiple learning paradigms. Notably, we find that contrary to their good usefulness under supervised learning, non-robust features attain poor usefulness when transferred to other self-supervised learning paradigms, such as contrastive learning, masked image modeling, and diffusion models. It reveals that non-robust features are not really as useful as robust or natural features that enjoy good transferability between these paradigms. Meanwhile, for robustness, we also show that naturally trained encoders from robust features are largely non-robust under AutoAttack. Our cross-paradigm examination suggests that the non-robust features are not really useful but more like paradigm-wise shortcuts, and robust features alone might be insufficient to attain reliable model robustness.

A multitude of explainability methods has been described to try to help users better understand how modern AI systems make decisions. However, most performance metrics developed to evaluate these methods have remained largely theoretical -- without much consideration for the human end-user. In particular, it is not yet clear (1) how useful current explainability methods are in real-world scenarios; and (2) whether current performance metrics accurately reflect the usefulness of explanation methods for the end user. To fill this gap, we conducted psychophysics experiments at scale ($n=1,150$) to evaluate the usefulness of representative attribution methods in three real-world scenarios. Our results demonstrate that the degree to which individual attribution methods help human participants better understand an AI system varies widely across these scenarios. This suggests the need to move beyond quantitative improvements of current attribution methods, towards the development of complementary approaches that provide qualitatively different sources of information to human end-users.

Abstract--In this letter, we investigate whether classical function allocation--the principle of assigning tasks to either a human or a machine--holds for physical Human-Robot Collaboration, which is important for providing insights for Industry 5.0 to guide how to best augment rather than replace workers. This study empirically tests the applicability of Fitts' List within physical Human-Robot Collaboration, by conducting a user study (N=26, within-subject design) to evaluate four distinct allocations of position/force control between human and robot in an abstract blending task. We hypothesize that the function in which humans control the position achieves better performance and receives higher user ratings. When allocating position control to the human and force control to the robot, compared to the opposite case, we observed a significant improvement in preventing overblending. This was also perceived better in terms of physical demand and overall system acceptance, while participants experienced greater autonomy, more engagement and less frustration. An interesting insight was that the supervisory role (when the robot controls both position and force) was rated second best in terms of subjective acceptance. Another surprising insight was that if position control was delegated to the robot, the participants perceived much lower autonomy than when the force control was delegated to the robot. These findings empirically support applying Fitts' principles to static function allocation for physical collaboration, while also revealing important nuanced user experience trade-offs, particularly regarding perceived autonomy when delegating position control. Received 7 May 2025; accepted 25 October 2025.