Artificial Inteligence

현재 인기있는 인공지능에 대한 Technical 관련된 수업을 건국대학교에서 듣고왔다.

강연자는 Intel출신 과학자로 Artificial Intelligence의 작동원리 같은것을 PPT로 소개해주셨다. 

Artificial intelligence 

• Natively Neuromorphic Architecture with no instruction set

NeuroMem computing alternative

• Today's common platforms
• The Neuromemon platform

The 8 pillars of neuromorphic pantheon

: up to nm of neurons -> increasing 

• Broadcast mode
• Deterministic search time
• Winner takes all
• Uncertain response
• Back probational aren't
• No fetch and decode of program Instruction
• Beyond biology

We don't have software. generation capability 가 필요하다. 

The merger of two old, but still in-fashion concepts

• Non linear classfier
• Hardwired parallel architecture

(hardware를 먼저 만들고 그 이후에 software를 만드는 것 같다. ) 

그래서 이 인공지능 중 NeuroMem은 회사인가? 

The Neurome Innovation

• Reactive memories
• Demonstrated High 

From data to Insight

• you can put system in analytics 
• It should be comparable 
• Twitter, and anlaytics. 

Data

• Usage Models, Level 1 

모든 Dtata들을 모아서 Vector로 바꾼후 NeuroMem을 통해 Neuro response를 만들어낸다.

그래서 이 NeuroMem은 어떤 알고리즘을 바탕으로 한 것인가? 

Neurom이 하는 동작들

• Identify
• Classification
• Anomaly detection
• Novelty detection
• Clustering
• Tracking
• Template matching (?) 

Application

• A companion processor to off-load the Vanlenumainn machines from pattern recognition tasks
• Cognitive Sensing
• convert sensor signals into insight data at the source
• Cognitive Storage
• Cognitive networking

The ZISC architecture timeline

• 1990 Nestor and Dassualt Aircraft starts 

Algorithms

Do you have algorithms which