Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

Cognitive Science > Chapter 7: The Network Approach > Flashcards

Chapter 7: The Network Approach Flashcards

You may prefer our related Brainscape-certified flashcards:
Biology 101 flashcards
1
Q

What is the Artificial Neural Network (ANN)?

A

A computer simulation of how populations of real neurons might perform some task

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is the Input Layer?

A

The first layer of the three-layer network that receives stimulus input and where stimulus is represented

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What is the Output Layer?

A

The third layer of a three-layer network. It generates a representation of a response based on inputs from hidden layer.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is the Hidden Layer?

A

A second layer of a three-layer network. This is where the input layer sends its signals. It performs intermediary processing.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What problems are ANNs good at?

A
  • Problems of Classification (pattern recognition, concept formation)
  • Control Problems (programming of robot movements)
  • Problems of Constraint Satisfaction (for ill-defined problems)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What are Serial Processors?

A

Processors that perform one computation at a time. The result of a particular computing unit can then serve as the input to a second computation. (e.g.: traditional PCs)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is Parallel Distributed Processing?

A

E.g. brain or ANNs. Large numbers of computing units perform their calculations in parallel.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is the Knowledge-Based Approach to problem-solving?

A

Conceptualizing the problem and its solution in terms of symbolic representations and transformations of these symbolic representations.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is the Behaviour-Based Approach to problem-solving?

A

Leaving the computational details up to the network itself and not paying much attention to symbolic representations or rules

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is a Distributed Representation?

A

Mental representations as patterns of activation among the network’s elements.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is a Local Representation?

A

Representing concepts via activity of a single node

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What do nodes represent?

A

Neurons or basic computing units

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What do links represent?

A

Connections between nodes

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

When does a node fire?

A

When the input is greater than or equal to the threshold value.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What do weights specify?

A

They specify the strength of the links between the nodes

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is the Basis function?

A

It specifies the amount of stimulation a given node receives. It sums up all of the inputs the node receives multiplied by the weights associated with the connection between processing units

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

What is the Activation function?

A

It maps the strength of the inputs a node receives onto its output.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What did McCullosh and Walter Pitts propose?

A

They were the first researchers to propose how biological networks might function. They assumed that each neuron had a binary output (sends out a signal or not) and whether or not a neuron would fire was determined by a threshold value.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

What is a Cell Assembly?

A

A small group of neurons that repeatedly stimulate each other.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

What is a Phase Sequence?

A

A group of connected cell assemblies that fire synchronously or nearly synchronously

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

What are Perceptrons created by Rosenblatt on 1958?

A

Perceptrons are neural nets designed to detect and recognize, store and use patterned information. They can learn from experience.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

What is the Error signal?

A

The difference between the actual and desired output.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

Describe the Generalized Delta Rule learning model (also known as back-propagation)

A

The network uses the error signal to modify the weights of the links. The modified weights allow the network to generate a response that is closer to the desired one. After repeated presentations of the stimulus in the presence of feedback, the network is able to produce the target response.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

What are three basic types of Network Dynamics?

A

Wh1. Convergent (a significant amount of activity at first which then slows down)

  1. Oscillatory (The weighs fluctuate periodically)
  2. Chaotic (The network’s activity varies in a chaotic fashion)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

What is the Loss Function?

A

The change in the error signal over the set of learning trials. It is a way of assessing whether the activity level of the network settles down too soon.

26
Q

What are Supervised Networks?

A

They are presented with target answers for each pattern they are given as input. The network “knows” what the right answer is in each training trial.

27
Q

What are Unsupervised Networks?

A

The network must determine the answer on its own, without the benefit of an answer.

28
Q

Single-layer networks

A

They have only one layer of nodes

29
Q

Multi-layer networks

A

They have multiple layers of nodes

30
Q

What is a Feed-Forward Network?

A

The flow of activation is in one direction only—forward; flow is from units in an input layer to units in other layers.

31
Q

What is a Recurrent Network?

A

Information can flow in two directions.

32
Q

What are Hopfield-Tank networks?

A

A type of supervised, single-layer, laterally connected networks. These networks are autoassociative—they are good at regenerating clean versions of patterns they have prior experience of when presented with noisy or incomplete versions of those patterns as input.

33
Q

What is the Kohonen network?

A

An example of an unsupervised, two-layer network. It’s also called a feature map as the Kohonen network is able to create a topological map or spatial representation of the features that are present in the stimulus input.

34
Q

What is the Adaptive Resonance Theory network (ART)?

A

An example of an unsupervised, multi-layer, recurrent network. It is able to classify input patterns and put them into different categories in the absence of a teacher

35
Q

What is Biological Plausibility?

A

The idea that artificial neural networks effectively represent and model characteristics of real-world brains.

  1. Artificial networks share general structural and functional correlates with biological networks.
  2. Artificial networks are capable of learning.
  3. They react to damage in the same way that human brains do)
36
Q

What is Graceful Degradation?

A

A gradual decrease in performance with increased

damage to the network.

37
Q

Interference

A

Refers to instances in which two sets of information that are similar in content interfere with one another. Small networks trained to learn large numbers of patterns show signs of interference: they have difficulty in distinguishing similar patterns

38
Q

Generalization

A

The ability to apply a learned rule to a novel situation.

39
Q

Weaknesses of ANNs

A
  1. It is not yet possible to simulate parallel processing of the magnitude of the brain’s one;
  2. Many networks show convergent dynamic while real neural networks are oscillatory and chaotic.
  3. Networks may have inadequate learning rules.
40
Q

Stability-plasticity dilemma

A

It states that a network should be plastic enough to store novel input patterns; at the same time it should be stable enough to prevent previously encoded patterns from being erased

41
Q

Catastrophic interference

A

Occurs in instances in which a network has learned to recognize a set of patterns and then is called upon to learn a new set. The learning of the new set modifies the weights of the network in such a way that the original set is forgotten

42
Q

Semantic networks

A

In semantic networks, each node has a specific meaning. They employ a local representation of concepts.

43
Q

Spreading activation

A

In semantic network models, a node’s activity can spread outward along links to activate other nodes. These nodes can then activate still others. Spreading activation is thought to underlie retrieval of information from long-term memory.

44
Q

Retrieval cues

A

A phenomenon in which an item related to one that was memorized can lead to successful recall.

45
Q

Priming

A

Occurs when the processing of a stimulus is facilitated by the prior presentation of a related stimulus.

46
Q

Hierarchical organization (in semantic networks)

A

Organization of concept nodes in different levels from the most abstract down to the most concrete

47
Q

Sentence verification experiment

A

A procedure in which participants judge the truth or falsehood of sentences by pushing one of 2 buttons.

48
Q

Superordinate category

A

The most abstract form of conceptual category organization that encompasses all examples of the concept.

49
Q

Subordinate category

A

The most concrete or specific form of conceptual category organization.

50
Q

Ordinate category

A

A level of conceptual category organization of moderate specificity

51
Q

Cognitive economy

A

The principle that concepts should not have to be coded more times than necessary.

52
Q

“isa” link

A

A link in a propositional network that represents the relationships of belonging (e.g.: a bird is an animal)

53
Q

“hasa” link

A

A link in a propositional network that represents property relationships (e.g.: a bird has a beak)

54
Q

Agent link

A

Specifies the subject of the preposition, the one performing some action

55
Q

Object link

A

Denotes the object or thing to which the action is directed

56
Q

Relation link

A

Characterizes the relation between agent and object

57
Q

Token nodes

A

Specific instances or items within a category (e.g.: Jobke is a token node of the dogs category)

58
Q

T.O.T. phenomenon

A

This the acronym for “tip of the tongue” in which one feels a familiarity with an item but cannot quite recall it.

59
Q

Guided Search

A

Recalling an event using intelligence and reasoning

60
Q

Reconstructive Memory

A

Recalling an item based on guided search

that also may be subject to bias by subsequent information

Cognitive Science (12 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions