Prompt-Tuning has become a popular paradigm to leverage large-scale pre-trained language models (PLMs) in relation extraction (RE). We show that the rich prior knowledge of relational facts in PLMs, while improving RE performance, may have unintended harmful effects when semantically distant from relation descriptions. We investigate this phenomenon from a causal view, revealing that the pre-trained knowledge is a confounder leading to spurious correlations between relation representations and labels. We thus propose to intervene on relation representations via backdoor adjustment conditioning on verbalizers, attaining an interventional prompt-tuning method with a multi-mask orthogonality regularization mechanism. Specifically, we exploit multiple mutually orthogonal mask representations, each of which is paired with an individual verbalizer, to approximate verbalizer sampling in the intervention implementation. Our method impressively achieves competitive performance on four widely used RE datasets and significantly alleviates the misleading impacts of prior factual knowledge in PLMs.

This study introduces a robust planning framework that utilizes a model predictive control (MPC) approach, enhanced by incorporating signal temporal logic (STL) specifications. This marks the first-ever study to apply STL-guided trajectory optimization for bipedal locomotion, specifically designed to handle both translational and orientational perturbations. Existing recovery strategies often struggle with reasoning complex task logic and evaluating locomotion robustness systematically, making them susceptible to failures caused by inappropriate recovery strategies or lack of robustness. To address these issues, we design an analytical stability metric for bipedal locomotion and quantify this metric using STL specifications, which guide the generation of recovery trajectories to achieve maximum robustness degree. To enable safe and computational-efficient crossed-leg maneuver, we design data-driven self-leg-collision constraints that are 1000 times faster than the traditional inverse-kinematics-based approach. Our framework outperforms a state-of-the-art locomotion controller, a standard MPC without STL, and a linear-temporal-logic-based planner in a high-fidelity dynamic simulation, especially in scenarios involving crossed-leg maneuvers. Additionally, the Cassie bipedal robot achieves robust performance under horizontal and orientational perturbations such as those observed in ship motions. These environments are validated in simulations and deployed on hardware. Furthermore, our proposed method demonstrates versatility on stepping stones and terrain-agnostic features on inclined terrains.

This study proposes a novel planning framework based on a model predictive control formulation that incorporates signal temporal logic (STL) specifications for task completion guarantees and robustness quantification. This marks the first-ever study to apply STL-guided trajectory optimization for bipedal locomotion push recovery, where the robot experiences unexpected disturbances. Existing recovery strategies often struggle with complex task logic reasoning and locomotion robustness evaluation, making them susceptible to failures due to inappropriate recovery strategies or insufficient robustness. To address this issue, the STL-guided framework generates optimal and safe recovery trajectories that simultaneously satisfy the task specification and maximize the locomotion robustness. Our framework outperforms a state-of-the-art locomotion controller in a high-fidelity dynamic simulation, especially in scenarios involving crossed-leg maneuvers. Furthermore, it demonstrates versatility in tasks such as locomotion on stepping stones, where the robot must select from a set of disjointed footholds to maneuver successfully.

Diffusion models have emerged as a promising approach for behavior cloning (BC), leveraging their exceptional ability to model multi-modal distributions. Diffusion policies (DP) have elevated BC performance to new heights, demonstrating robust efficacy across diverse tasks, coupled with their inherent flexibility and ease of implementation. Despite the increasing adoption of Diffusion Policies (DP) as a foundation for policy generation, the critical issue of safety remains largely unexplored. While previous attempts have targeted deep policy networks, DP used diffusion models as the policy network, making it ineffective to be attacked using previous methods because of its chained structure and randomness injected. In this paper, we undertake a comprehensive examination of DP safety concerns by introducing adversarial scenarios, encompassing offline and online attacks, global and patch-based attacks. We propose DP-Attacker, a suite of algorithms that can craft effective adversarial attacks across all aforementioned scenarios. We conduct attacks on pre-trained diffusion policies across various manipulation tasks. Through extensive experiments, we demonstrate that DP-Attacker has the capability to significantly decrease the success rate of DP for all scenarios. Particularly in offline scenarios, we exhibit the generation of highly transferable perturbations applicable to all frames. Furthermore, we illustrate the creation of adversarial physical patches that, when applied to the environment, effectively deceive the model. Video results are put in: https://sites.google.com/view/dp-attacker-videos/.

News recommendation is of vital importance to alleviating in-formation overload. Recent research shows that precise modeling of news content and user interests become critical for news rec-ommendation. Existing methods usually utilize information such as news title, abstract, entities to predict Click Through Rate(CTR) or add some auxiliary tasks to a multi-task learning framework. However, none of them directly consider predicted news popularity and the degree of users' attention to popular news into the CTR prediction results. Meanwhile, multiple inter-ests may arise throughout users' browsing history. Thus it is hard to represent user interests via a single user vector. In this paper, we propose PENR, a Popularity-Enhanced News Recommenda-tion method, which integrates popularity prediction task to im-prove the performance of the news encoder. News popularity score is predicted and added to the final CTR, while news popu-larity is utilized to model the degree of users' tendency to follow hot news. Moreover, user interests are modeled from different perspectives via a subspace projection method that assembles the browsing history to multiple subspaces. In this way, we capture users' multi-view interest representations. Experiments on a real-world dataset validate the effectiveness of our PENR approach.