Graph Neural Networks (GNNs) have significantly boosted the performance of many graph-based applications, yet they serve as black-box models. To understand how GNNs make decisions, explainability techniques have been extensively studied. While the majority of existing methods focus on local explainability, we propose DAG-Explainer in this work aiming for global explainability. Specifically, we observe three properties of superior explanations for a pretrained GNN: they should be highly recognized by the model, compliant with the data distribution and discriminative among all the classes. The first property entails an explanation to be faithful to the model, as the other two require the explanation to be convincing regarding the data distribution. Guided by these properties, we design metrics to quantify the quality of each single explanation and formulate the problem of finding data-aware global explanations for a pretrained GNN as an optimizing problem. We prove that the problem is NP-hard and adopt a randomized greedy algorithm to find a near optimal solution. Furthermore, we derive an improved bound of the approximation algorithm in our problem over the state-of-the-art (SOTA) best. Experimental results show that DAG-Explainer can efficiently produce meaningful and trustworthy explanations while preserving comparable quantitative evaluation results to the SOTA methods.

Deep graph networks (DGNs) have demonstrated their outstanding effectiveness on both heterogeneous and homogeneous graphs. However their black-box nature does not allow human users to understand their working mechanisms. Recently, extensive efforts have been devoted to explaining DGNs' prediction, yet heterogeneity-agnostic multi-level explainability is still less explored. Since the two types of graphs are both irreplaceable in real-life applications, having a more general and end-to-end explainer becomes a natural and inevitable choice. In the meantime, feature-level explanation is often ignored by existing techniques, while topological-level explanation alone can be incomplete and deceptive. Thus, we propose a heterogeneity-agnostic multi-level explainer in this paper, named HENCE-X, which is a causality-guided method that can capture the non-linear dependencies of model behavior on the input using conditional probabilities. We theoretically prove that HENCE-X is guaranteed to find the Markov blanket of the explained prediction, meaning that all information that the prediction is dependent on is identified. Experiments on three real-world datasets show that HENCE-X outperforms state-of-the-art (SOTA) methods in generating faithful factual and counterfactual explanations of DGNs.

This paper presents the mechatronic design and experimental validation of a novel powered knee-ankle orthosis for testing torque-driven rehabilitation control strategies. The modular actuator of the orthosis is designed with a torque dense motor and a custom low-ratio transmission (24:1) to provide mechanical transparency to the user, allowing them to actively contribute to their joint kinematics during gait training. The 4.88 kg orthosis utilizes frameless components and light materials, such as aluminum alloy and carbon fiber, to reduce its mass. A human subject experiment demonstrates accurate torque control with high output torque during stance and low backdrive torque during swing at fast walking speeds. This work shows that backdrivability, precise torque control, high torque output, and light weight can be achieved in a powered orthosis without the high cost and complexity of variable transmissions, clutches, and/or series elastic components.

Traditional control methodologies of rehabilitation orthoses/exoskeletons aim at replicating normal kinematics and thus fall into the category of kinematic control. This control paradigm depends on pre-defined reference trajectories, which can be difficult to adjust between different locomotor tasks and human subjects. An alternative control category, kinetic control, enforces kinetic goals (e.g., torques or energy) instead of kinematic trajectories, which could provide a flexible learning environment for the user while freeing up therapists to make corrections. We propose that the theory of underactuated potential energy shaping, which falls into the category of kinetic control, could be used to generate virtual body-weight support for stroke gait rehabilitation. After deriving the nonlinear control law and simulating it on a human-like biped model, we implemented this controller on a powered ankle-foot orthosis that was designed specifically for testing torque control strategies. Experimental results with an able-bodied human subject demonstrate the feasibility of the control approach for both positive and negative virtual body-weight augmentation.