4 accepted papers
Current adversarial examples (AEs) are typically designed for static models. However, with the wide application of Class-Incremental Learning (CIL), models are no longer static and need to be updated with new data distributed and labeled differently from the old ones. As a result, existing AEs often
Substitute training-based data-free black-box attacks pose a significant threat to enterprise-deployed models. These attacks use a generator to synthesize data and query APIs, then train a substitute model to approximate the target model's decision boundary based on the returned results. However, ex…
This letter presents a lockable upper-limb exoskeleton to reduce the physical fatigue of laborers/surgeons operating with upper limbs maintaining fixed posture for long time. A novel structure design integrating the jamming and engagement mechanisms that can provide locking of flexion and adduction
In wearable applications, traditional pneumatic artificial muscles (PAMs) cannot fully satisfy the requirements of safe and sufficient interaction with the human body. The requirements include high contraction ratio, self-contained sensing, reconfiguration, locking abilities, and avoidance of the sq