Development of Virtual Reality (VR) applications is no doubt gaining momentum and is soon to be the next big thing, similar to what we saw with personal computers and smartphones. One that has already began to gain momentum is the field of VR based training or stroke rehabilitation research. For almost a decade VR based rehabilitation research has shown promising results, indicating that the training received using VR is comparable with that of conventional rehab methods. The main idea is to interface therapist recommended rehabilitation tool or training equipment with VR scenarios and then control the VR scenario by performing a particular exercise. Even though with such fast pace development of VR based rehabilitation tools, one question that still remains to be answered is the availability of a platform that can allow therapist to carry out rehabilitation research related to both upper and lower extremity . This research work try to answer the problem by introducing a multi-interface platform that can be used for both upper and lower extremity rehabilitation and training research. The entire platform is developed in such a way that the VR and interfaces maintain a two way communication link, i.e. the interfaces can control VR scenarios and the VR scenarios can control the interface working behavior. Besides integration process of the device with the VR scenario, the study also proposes several signal processing and machine learning techniques so as to ensure that therapist recommended exercises are properly performed. A teaching method known as 'operant conditioning' is implemented to ensure that the subject performs exercises correctly. Apart from implementing multi-interface training platform, the study also outlines a method to detect the progress of the patient’s performance and produce the output in terms of Berg Scale Ratings. Besides the rehabilitation and training research the study also describes a compressive sensing technique to study the effect on various machine learning technique in scenario recognition task.