Imitation Learning: Progress, Taxonomies and Challenges.

Imitation learning (IL) aims to extract knowledge from human experts' demonstrations or artificially created agents to replicate their behaviors. It promotes interdisciplinary communication and real-world automation applications. However, the process of replicating behaviors still exhibits various problems, such as the performance is highly dependent on the demonstration quality, and most trained agents are limited to perform well in task-specific environments. In this survey, we provide an insightful review on IL. We first introduce the background knowledge from development history and preliminaries, followed by presenting different taxonomies within IL and key milestones of the field. We then detail challenges in learning strategies and present research opportunities with learning policy from suboptimal demonstration, voice instructions, and other associated optimization schemes.

Attn-HybridNet: Improving Discriminability of Hybrid Features With Attention Fusion.

The principal component analysis network (PCANet) is an unsupervised deep network, utilizing principal components as convolution filters in its layers. Albeit powerful, the PCANet suffers from two fundamental problems responsible for its performance degradation. First, the principal components transform the data as column vectors (which we call the amalgamated view) and incur a loss of spatial information present in the data. Second, the generalized pooling in the PCANet is unable to incorporate spatial statistics of the natural images, and it also induces redundancy among the features. In this research, we first propose a tensor-factorization-based deep network called the tensor factorization network (TFNet). The TFNet extracts features by preserving the spatial view of the data (which we call the minutiae view). We then proposed HybridNet, which simultaneously extracts information with the two views of the data since their integration can improve the performance of classification systems. Finally, to alleviate the feature redundancy among hybrid features, we propose Attn-HybridNet to perform attention-based feature selection and fusion to improve their discriminability. Classification results on multiple real-world datasets using features extracted by our proposed Attn-HybridNet achieves significantly better performance over other popular baseline methods, demonstrating the effectiveness of the proposed techniques.

A Low-Cost Adaptive Balance Training Platform for Stroke Patients: A Usability Study.

Stroke patients usually suffer from asymmetric posture due to hemi-paresis that can result in reduced postural controllability leading to a balance deficit. This deficit increases the risk of falls, which often makes them dependent on caregivers for community ambulation, thus deteriorating their quality of life. Conventional balance training involves rehabilitation exercises performed under physiotherapist's supervision, where the scarcity of trained professionals as well as the cost of clinic-based rehabilitation programs can deter stroke survivors from undergoing regular balance training. Thus, researchers have been exploring technology-assisted solutions, e.g., home-based virtual reality (VR) setup. In this paper, we developed a VR-based balance training (VBaT) platform, where VR-augmented user-interface using Nintendo Wii balance boardwas tested in a laboratory setting for its feasibility. The VBaT offered tasks of varying difficulties to the participants that adapted to individual performance capability during balance training. We performed a preliminaryusability study with 7 stroke survivors (post-stroke period > 6 months). Preliminary results indicate the potential of theVBaT system to cause improvement in overall average task performance over the course of training while using the VBaT. Thus the VBaT system is proposed to be a step toward an effective balance training platform for people with balance disorder.

Audio-Visual Stimulation in Conjunction with Functional Electrical Stimulation to Address Upper Limb and Lower Limb Movement Disorder.

Neurological disorders often manifest themselves in the form of movement deficit on the part of the patient. Conventional rehabilitation often used to address these deficits, though powerful are often monotonous in nature. Adequate audio-visual stimulation can prove to be motivational. In the research presented here we indicate the applicability of audio-visual stimulation to rehabilitation exercises to address at least some of the movement deficits for upper and lower limbs. Added to the audio-visual stimulation, we also use Functional Electrical Stimulation (FES). In our presented research we also show the applicability of FES in conjunction with audio-visual stimulation delivered through VR-based platform for grasping skills of patients with movement disorder.