Top Banner
The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University
39

The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Dec 18, 2015

Download

Documents

Mariah Brown
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The Gold Seekers Project

Crescent Lab

Department of Computer Science

Texas Christian University

Page 2: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Three Sub-Projects Alchemy

A distributed process environment for AI applications and cognitive models

Can run on a cluster of a grid Gold Mind

Cognitive modeling environment based on Alchemy El Dorado

Improved distributed AI processing theory Software/Hardware applications of Alchemy/Gold

Mind

Page 3: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Relationship of the Projects

C og nitionC oncepts

B rain/MindP roces s ingC oncepts

Improved Cog nitiveMode l ing Environment

D is tr ibutedP roces s ingConcepts

Improved S pec ial-P urpos eDis tr ibuted Environment

New Concepts in(Geog raphical ly)

D is tr ibuted Computing

G o a l M ind A lc he m y E l D o ra do

UsefulApplications

Page 4: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Alchemy

Page 5: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Geographically Distributed AI System Requirements

Asynchronous processing nodes most inference engines use some type of

constantly repeating match-select-fire cycle

Dynamic collections of nodes dividing an application’s problem space

into sub-spaces requires a graph of reasoners which share a flexible communication protocol

Implementation needs to exhibit: good overall processing times speedup secure communication

Page 6: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The Alchemy Approach Build on other distributed architectures:

CORBA, IBM Aglets & Sun’s JINI AMEBA (Hannon and Cook, 2001)

Create a GDAI friendly: Threading model (housekeeping, clients, servers) Security model (authentication, encryption, tracking) Development environment (GUI-based)

Implement it using an OOD in C++ Test using different applications

Formally-defined web-based search of documents

Page 7: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Alchemy Components

SSL

Alchemy Security Model

Node

Handler

Client Server

Housekeeper

Alchemy Client-Server

Model

GADGET

ProcessControl

MigrationControl

ApplicationNodes

ThreadingModel

Page 8: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Alchemy Architecture

Processor Node Processor Node Processor Node Processor Node

GADGET Migration Control

Client-Server Connections

Process Control Process Control Process Control Process Control

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

Page 9: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Alchemy Resources The Crescent Lab Beowulf Cluster

Housed on the third floor TTC 12x2 + 7x1 processor array

28.4 GHz of total processing power

Uses 3 – switched 100baseT networks Effective bandwidth of 600 Mb/sec

Page 10: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Handling Message Security Authentication

complex response embedded time stamp first key (public, super secure, pre-session)

Encryption algorithm (strong or weak) public-private vs. symmetric keys key and sieve length

Message tracking digesting approach

Page 11: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Alchemy’s Approach Embedded support

C++ class libraries based on SSL Multi-level

support layer’s level is locked by system application node level determined by server

Trade-off between speed and safety 256 bit Blowfish MD5

Page 12: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Handling Network Security Common port mapping

e.g., a local port to 8080 to local port requires some kind of server interface and the application

intelligence to know what to route VPN

to the inside network elements looks like any other network device has a IP address, mask and route, etc.

to each other looks like a peer-to-peer (UDP) or client-server (TCP) connection tunnels messages over a [secure] pipe

routes anything that knows the VPN’s address

Page 13: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Handling Network Security (2) Intelligent Private Network (IPN)

Piggybacks on the Alchemy server To simplify application design, Alchemy uses name spaces

the application server’s port is maintained by the local Alchemy server

application clients always ask for server connections by name, never by host and port

the this name is translated into a host and port by the client’s local Alchemy server and returned to the client via the support system

Using IPN, when an application’s client ask for remote server, it gets the host and port of the local IPN

Page 14: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

(Picture of Beowulf here)

Page 15: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Gold Mind

Page 16: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The Gold Mind Mission Develop a better understanding of human

cognition and interaction by modeling human performance on a computer system

Support the creation of individual models of humans performance on a given task which can be later be integrated into larger models of more complicated tasks

In the far distance future, solve the ‘grand challenge’ problem of a generalized model of human cognitions

Page 17: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The Gold Mind Architecture

Page 18: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The Etheron Computational Model

Control

To ControlDaemonView/

Edit

Trace

SystemParent

AgentParent

Process/Inference

Ch

ild

ren

To Knowledge Tool

To Trace Tool

To System(a parentEtheron)

To Agent(a parentEtheron)

Listen forconnections

To other Etherons

Page 19: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Current Gold Mind Models TALLUS

Explores language use and learning in young children STRESS

Explores a cognitive model for the Stroop effect FAME

Explores a filter-fuser mechanism of attention and arousal

ED-FAME Explores how emotions control for attention and arousal

Page 20: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The TALLUS Model Teacher Assisted Language Learning and

Understanding Simulation Contains three or more agents (one teacher and two

or more students) Teacher is simply a HMI for entering adult level

speech and seeing the telegraphic output Students have the same computational and

cognitive design, but different knowledge

Page 21: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The TALLUS Model Overview

Page 22: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

One of the Model’s Student Agents

S oc ia l K now ledge B as eE go K nowledge B as e

Convers at ion K nowledge B as e Dis c ours e K nowledge B as e

HearingInterfac e

S y ntac t ic C las s ifier Lex ic on C las s ifierS em ant ic s Reas oner

S peak ingInterfac e

V is ion C las s ifier

E pis odic Reas oner S em ant ic Conc ept Reas oner

V is ion S t im uli Router

Agent

Router

V is ionInterfac e

S urfac e S truc ture G enerator

Utteranc e S t im uli Router

S peec h S t im uli Router

K nowledge S t im uli Router

H igher O rder P roc es s es S t im uli Router

Page 23: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Discourse Example from TALLUS

Teacher: Jill, what is this? (holding a blue ball)Jack: A ball.Teacher: Jack, let Jill answer. (pause) Jill, what is it?Jill: A ball.Teacher: Yes, that is right! It is a ball. What color is it?Jill: Blue ball.Teacher: Yes, that is right! It is a blue ball.

Jack, what do you do with it?Jack: Play ball.Teacher: Yes, that is right! You can play with a ball.

Page 24: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

TALLUS Speedup Results

1

2

3

4

5

6

1 2 3 4 5

Number of Nodes

Sp

eed

up

100MB/sec

500MB/sec

Linear

Page 25: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The STRESS Model

Stroop Test Response Evaluation Sub-System

Contains three agents (one researcher, one evaluation and one subject) Researcher agent is simply a HMI for entering testing

commands and viewing the subject’s responses Evaluation agent generates test based on input from a

HMI Subject agent is similar to a TALLUS student

Page 26: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The STRESS Model

R e s e ar c he rA g e nt

C om mHub

S ubje xtA g e nt

Le xic alA nalyz e r

V is ionS R N

HO PS R N

S e m antic sR e as one r

Tas kR e as one r

C olorA n alyz e r

P atte r nA n alyz e r

Knowledg eS RN

ConceptReas oner

Utte r anc eP r oc e s s ing

S R N

Utte r anc eGe ne r ation

S R N

V is ionI/F

Lang uag e /Tim e S ync

I/F

EvaluationA g ent

HMI

S tr oopTe s t

Ge ne r ator

Tim e S yncI/F

Utte r anc eGe ne r ation

R e as one r

Lang uag eI/F

Lang uag eI/F

V is ionI/F

Tim e S yncI/F

S yntaxA nalyz e r

Page 27: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The STRESS Results

400

500

600

700

800

900

Congruent Control Conflict

Rea

ctio

n ti

me

(mil

lise

cond

s)

Word Reading

Color Naming

40

50

60

70

80

90

100

110

120

130

Congruent Control Conflict

Rea

ctio

n ti

me

(mil

lise

cond

s)

Word Reading

Color Naming

HUMAN STRESS

Page 28: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The FAME Model Attention and arousal function divided into

two functions – filtering and fusion three aspects – time, space and modality Resulting in 3 component types

Time Filter/Fuser () Space Filter/Fuser () Mixed Modality Filter/Fuser ()

Allows these component to be combined as a graph (most often a lattice) with sensor input and processing elements

Page 29: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The FAME Application

Room

Temp #1

Temp #2

Light #1

Light #2

AttentionControl

SensorProcessing

ArousalControl

Page 30: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The ED-FAME Application Model

Page 31: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

El Dorado

Page 32: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Current El Dorado Projects PRISE

Integration of a mobile robot and room-based sensors in a smart home environment to provide a mobile assistant.

LiMM Multi-terrain sentry and inspection robot based on

biologically inspired design concepts.

RoAMS Track-based arm assistant for assisted living

Page 33: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Crescent Beowulf Cluster

PRISE Environment

Mobile Sensorand Tools Platform

FixedSensors

Wireless Network

Alchemy

Gold Mind Application

Page 34: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Crescent Beowulf Cluster

LiMM Environment

Sensor, Legs and Manipulators LAN/WAN Wireless

Network

Alchemy

Gold Mind Application

Alchemy

Gold Mind Application

The LiMM Controller Network

Page 35: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The LiMM

Page 36: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

The LiMM Robot Design Plan Phase I

Build the robot platform and six legs Uses 7 on-board µController for leg control Power supplied by a tether

Phase II Add end manipulators to the legs Uses same set of on-board µControllers for leg

and manipulator control Still uses tethered power

Page 37: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

LiMM Design Plan (2) Phase III

Add a sensor suite with its own set of µControllers

Add an on-board power supply Phase IV

Add an on-board µProcessor for remote operations

Add a WAN network capability using a satellite link or something like GSM

Page 38: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

LiMM (Phase I) Facts Each of the six legs contains

5 DC motors A leg control module contain a rabbit µC

A Ethernet network connects the six leg controllers to both a master controller the the wireless network

Page 39: The Gold Seekers Project Crescent Lab Department of Computer Science Texas Christian University.

Crescent Beowulf Cluster

RoAMS Environment

AlchemyAlchemy

Gold Mind Application

The LiMM Controller Network

FixedSensors

Robot Arm and Track