For Project Enquiry +91 9962 067 067

EXISTING SYSTEM

PROPOSED SYSTEM

EXISTING CONCEPT:-

In the existing system various approaches to integrate domain knowledge and different types of learning systems have been known. In many of these works, domain knowledge is used in an indirect manner.

Bayesian networks, on the other hand, have used domain knowledge in both direct and indirect ways. Direct use of domain knowledge in Bayesian networks includes, where domain knowledge expressed.

PROPOSED CONCEPT:-

In the proposed system BMI user needs to learn how to operate a BMI system with biofeedback, e.g., visual feedback. They can only adjust their brain activities and observe the predicted movements executed by the external robot arm in reaching the target through trial and error.

RL theory becomes such an approach applied on the co-adaptive BMI studies.

EXISTING ALGORITHM:-

Q-learning techniques Algorithm

PROPOSED TECHNIQUE:-

AGREL

ALGORITHM  DEFNITION:-

In such scenarios, the rich movements and the corresponding high dimensional neural states (recording from monkey usually provides more channels of spikes than rats), form a large state-action space, which imposes curse of dimensionality on Q-learning techniques

TECHNIQUE DEFNITION:-

In our work global signal that is delivered to all units regardless of whether they were involved in the network’s choice. The second factor is the attention feedback signal from units in the output layer. This factor gates the plasticity of units at earlier processing levels responsible for the network’s output.

DRAWBACKS:-

Domain knowledge is typically used indirectly in RL systems.

FALCON is compatible with symbolic rule-based representation

ADVANTAGES:-

Improve the performance of BMI systems

AGREL is applied to map the consecutive movements