Please enable JavaScript.
Coggle requires JavaScript to display documents.
A survey on applications and security of embedded machine learning for…
A survey on applications and security of embedded machine
learning for smart healthcare systems
1.Introduction
The necessaries of using TinyML in Healthcare
the security and privacy of the patient's sensitive data, information
Storage of a huge volume of raw data
The computing resource at the cloud for traditional ML based system
expensive and power- hungry Graphics Processing Units (GPUs)
latency increasing, it is critical issue in some cases
the slightest delay maybe life-threatening
The research gap and motivation
Most of works:
the important of TinyML in healthcare systems
present the applications of TinyML
Present the limitations of TinyML
Process Power
Limited Memory
security and privacy (very limited)
Present the advantages of TinyML
Gaps:
security of deploying TinyML
Applications of Combined of TinyML and Opensource hardware (such as RISC-V)
those devices share many energy and com- putational power constraints
to implement encryptions schemes and other security mechanisms
Define of healthcare system
2. Embedded machine learning for smart
healthcare devices
3. Vulnerabilities of TinyML in smart
healthcare devices
reverse engineer the neural network architecture
Privacy of sensitive data
Hardware trojan due to the open source hardware
Authentication of Edge devices
EML frame work support RISC-V?
4. Security solutions for embedded machine
learning based smart healthcare systems
5. Conclusion