Connectionism PSYCH 85
Popular in Introduction to Cognitive Science
verified elite notetaker
Popular in Psychlogy
This 2 page Class Notes was uploaded by Morgan Hawes on Monday October 26, 2015. The Class Notes belongs to PSYCH 85 at University of California - Los Angeles taught by Kellman in Fall 2015. Since its upload, it has received 26 views. For similar materials see Introduction to Cognitive Science in Psychlogy at University of California - Los Angeles.
Reviews for Connectionism
Eugh...this class is soo hard! I'm so glad that you'll be posting notes for this class
-Leone Bechtelar Jr.
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 10/26/15
Cog Sci 102015 What is the main idea of each reading If you see the material presented in more than one way the better it is for your memory 1 Symbol processing and alternative proposal in a connectionist network you do not have symbols you have connections 2 Take home message connectionism neuralnetworks and parallel distributied processing this represents another way to do information processing 3 Neural networks get rid of symbol processing This is a good canidate as to how we get lots of stuff This is a good way to connect the hardware with the software Marr s 3 levels a To understand information processing pahominal you have to understand 3 levels i Ecological ii Level of representation and algarith iii Biological mechanism or hardware 6 Sometimes connectionist argue that you blur the lines between agerism and hardware a So it is not that the hardware and software are the same but you still need a process description 7 They use computers to simulate neural networks so it does not get rid of marr39s levels However this program is very different from other programs 8 208 these are different programs then the symbol system 9 There is no place to store something like your name in a connectionist network it could give you an approximate name but not your name this requires a symbol storage system It is important to take away a mixed architecuture prespective between connectionism and symbol systems 10We say neural networks are these real neurons No these are just modeled after the way neurons work in our system Those who study neural networking study abstrace neurons 11Neura networks are the way your brain works side because these send messages just like real neurons Each neuron has about 1000 conections symbolic processing is more insoated than this just one connection for storage systems a If you look a dog running though the room how do you know it is a dog You could not give an algorithim explain why it is a dog but only a connection between an image and a dog 12Neura networks are not the way your brain works side a Well they are abstract neurons not real neurons b Most of these models are using back progporgation where this a weight that goes back to change c Back propagation and Supervised learning look at ouput on network and see error signal and then the message gets sent to correct error min 42 91 i These are not how neurons works ii This look like behaviorism language disproves this d Math is too precise to operate on this model 13Speci cs of reading a Figure 81 b This is not symbol processing this is passing activation of one unit to another c There is a stimulus which has stripes or larger than car or wings or spots no legs and each one of these nodes makes a coneetoin and the out put is tiger bird snake leopard dog And each of these connections have a weight and these weights point us in the right connchon d In the beginning the weights are random and we have to retrain through feed back There are no rules or algorithims e Pg 216 neural networks are good at Notes for of ce hours Psychiatry is frued and drugs Andy christainson Bruce baker clinical psychologists rena repetti Clinical psychology is harder then med school Thomas bradbuy robles repetti
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'