Edge Artificial Intelligence Seminar Presentation.pptx

seminar presentation Edge AI

1. Edge AI Bringing Intelligence to the Edge Dr. UMA.S Assistant Professor, Department of Computer Science, Government Arts and Science College – Tirupattur

2. AGENDA Introduction to Edge AI Key Characteristics of Edge AI Key Components of Edge AI Advantages of Edge AI Applications of Edge AI Challenges Future Trends in Edge AI

3. Introduction Edge AI , or Edge Artificial Intelligence , D eployment of artificial intelligence (AI) algorithms and models directly on edge computing devices , rather than relying on centralized cloud servers for processing. Edge AI brings the power of AI and machine learning closer to the data source , enabling real-time data analysis and decision-making at the edge of the network, near the device or sensor producing the data.

4. Ability to overcome challenges associated with traditional centralized computing approaches. P lays a pivotal role in addressing the evolving demands of modern technology by offering low-latency , secure, and efficient processing at the edge of the network

5. Key Characteristics of Edge AI Decentralized Processing Real-time Decision-Making Data Privacy and Security Bandwidth Efficiency Offline Functionality Reduced Dependency on Cloud Services Distributed Architecture Optimized Resource Utilization

6. Edge AI and Cloud-based AI Edge AI and Cloud-based AI T he choice between them depends on the specific requirements of the application. Have their strengths and weaknesses Edge AI W ell-suited for scenarios demanding low latency , privacy , and offline functionality C loud-based AI E xcels in applications with large-scale data processing , centralized management , and scalability.

7. Edge AI Cloud-based AI Architecture Processing occurs at Local device Processing occurs at centralized servers Latency Lower Higher Bandwidth Usage Lower bandwidth usage Higher bandwidth costs Privacy and Security Security risks are mitigated Risk involved

8. Edge AI Cloud-based AI Offline Functionality Independent of continuous internet connectivity Continuous internet connection Scalability Provides scalability without overwhelming a central server. Offers centralized scalability Customization and Personalization Localized Centralized Resource Utilization Optimizes resource utilization Robust cloud infrastructure

9. Key Components of Edge AI Edge Devices S martphones, IoT devices, edge servers AI Algorithms Algorithms optimized for edge computing Edge Connectivity Communication protocols and Connectivity options

10. Edge Devices A n integral part of edge computing C omputing devices that perform data processing and storage closer to the location where the data is generated , rather than relying on a centralized cloud server. P aradigm that aims to bring computational capabilities closer to the source of data , reducing latency and enabling real-time processing.

11. Diverse and encompass a wide range of technologies, from smartphones and IoT devices to edge servers and embedded systems . C ollectively contribute to the evolution of edge computing, enabling decentralized processing , reducing latency , and supporting a variety of applications across different industries.

12. Types of Edge Devices Smartphones IoT Devices Edge Servers Gateways Embedded Systems Drones and Autonomous Vehicles Wearables

13. AI Algorithms Optimizing AI algorithms for edge computing involves tailoring models and methods to run efficiently on resource-constrained devices. When optimizing AI algorithms for edge computing, the emphasis is on reducing the model size , computational complexity , and memory requirements while maintaining satisfactory performance.

14. These optimizations are crucial for ensuring that AI applications can run efficiently on edge devices with limited resources .

15. AI Algorithms MobileNet TinyML EdgeTPU Models TensorFlow Lite ONNX Runtime Bonsai Apache OpenNLP Kaldi

16. Edge connectivity Edge connectivity is a critical aspect of Edge AI, I t involves the communication between edge devices and the broader network . S electing the appropriate communication protocols and connectivity options for Edge AI depends T he specific requirements of the application, F actors like latency , bandwidth , power consumption , and the nature of the data being exchanged.

17. The goal is to establish a robust and efficient communication infrastructure that supports the real-time processing demands of edge devices.

18. C ommunication protocols MQTT (Message Queuing Telemetry Transport) CoAP (Constrained Application Protocol) HTTP/HTTPS DDS (Data Distribution Service) AMQP (Advanced Message Queuing Protocol)

19. Connectivity Options 5G Networks Edge Servers and Cloudlet Computing Fog Computing Mesh Networks LPWAN (Low-Power Wide-Area Network

20. Advantages of Edge AI Low Latency Edge AI reduces latency compared to cloud-based solutions Privacy and Security Emphasizing local processing for enhanced privacy Bandwidth Efficiency Reducing the need for continuous data transfer

21. Applications of Edge AI Smart Cities Applications in traffic management, public safety, etc. Healthcare Remote patient monitoring, predictive analytics Manufacturing Quality control, predictive maintenance Retail Personalized customer experiences, inventory management

22. Challenges Limited Resources Dealing with constrained computing power and storage Security Concerns Addressing potential security risks Standardization The need for standardized frameworks in Edge AI

23. Future Trends in Edge AI 5G Integration 5G will enhance Edge AI capabilities Continued Miniaturization Advancements in hardware for smaller , more powerful edge devices Enhanced Edge-to-Cloud Collaboration Tasks dynamically shift between the edge and the cloud based on processing requirements

24. Text generated by ChatGPT , November 23, 2023, OpenAI , https://chat.openai.com .

25. THANK YOU