Spime - personal assistant
Download as PPT, PDF•0 likes•1,280 views
This document proposes a personal assistant called a "Spime" that is based on IoT technology. The Spime would have decision-making and self-learning abilities, and the ability to interact with other Spimes. It describes the Spime's system structure using various sensors that collect data which is transferred to the cloud for computing and storage. The document outlines the procedural flows among components, including how the Spime would encrypt data transferred to the cloud. It provides examples of how the Spime could be used to provide basic information to users, recognize objects and locations, and allow the cloud to perform more advanced tasks like data mining for recommendations. Privacy controls are also discussed.
Spime - personal assistant
- 1. Spime: Personal Assistant 995002513 Kir Chou 1
- 2. Define Based on IOT Additional Decision-making ability Self-learning ability The ability of interacting with other spimes 2
- 5. System structure with components 5 Cloud Wireless Module Sensor Processor Thermometer Gyro Accelerometer Magnetometer Luminance meter Manometer Hygrometer Microphone Speaker …etc S p i m e Data transfer Others cache
- 6. Procedural flows among components 6 Wireless Module Because of space, energy restrictions, ⇒reduce the amount of calculation and storage To reduce the amount of calculation ⇒Transfer data to cloud for computing and store Cloud 1. Permanent Connection 2. Exchange Data
- 7. Procedural flows among components 7 Catch all possible physical data ⇒Environment situation, Biometrics, Voice… Communication with users ⇒Voice Sensor Voice
- 8. Procedural flows among components 8 Encrypt and decrypt data between cloud and spimes ⇒To avoid MITM or other attacks Wireless Module Sensor Processor cache Cloud C C P P C C C P C P * 存非 式加密 制現 對稱 機 *Cache and store temporary data
- 9. Interact with others 9 1. After user agree Send a special pattern to other spimes 2. Then cloud can access other spimes’ data A B
- 10. Use case / Actions/ Operations 10 Passive: User ask question (higher privacy) ⇒Spime reply user Active: User doesn’t ask question ⇒Spime decide whether to provide data to user or not (Based on machine learning result)
- 11. Use case / Actions/ Operations Basic (higher privacy) : search Weather data Transportation data Environmental recognition Advanced : store Line tracing Spime recognition(meet views, area, time) 11 SRC SRC
- 12. Use case / Actions/ Operations Cloud data mining (lower privacy) : manage Schedule reminder Life suggestion(eat, wear, transportation) Environmental prediction Spime prediction Healthy management Early learning Chatting Agency identity response Explore new demand 12 SRC Not yet!
- 13. User interface with your preference style Privacy mechanism Role: Group / Field / Friend Service: The level of machine learning 13 http://select.cs.cmu.edu/code/graphlab/images/ppt/fig_10.gif
- 14. Why we don’t have the kind of things? Privacy responsibility Limitation of Energy Limitation of AI 14 *According from my startup friend.
- 15. Thank you 15