R17 Questions Flashcards

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

What major accomplishment is achieved by the Deep Mind software described by Koch?

A

Deep mind software developed its own strategies to beat human players in Breakthrough.

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

What 3 features from neurobiology are incorporated in the algorithm?

A

Deep convolution networks, reinforcement learning (operant conditioning), & selective memory replay

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

At the beginning of training, the algorithm improves based on _______________.

A

trial and error

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

What is a deep convolutional neural network (CNN)?

A

It uses multiple layers of simulated neurons that have weighted inputs. Units turn on and off depending on the sum of input weights they receive.

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

CNNs are modeled after which brain system?

A

The mammalian visual system

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is the difference between supervised learning and reinforcement learning?

A

In supervised learning, every input image is paired with a specific label. In reinforcement learning, the consequence of any action in a game score unfolds in time.

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

What is selective memory replay and where in the brain is it thought to occur?

A

Selective memory replay is when neurons repeat the same patterns of activation displayed earlier.

It occurs in the hippocampus.

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

Name one weaknesses of the algorithm?

A

One weakness is that CNNs are unable to plan for the long-term.

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