Most Popular

1500 questions
7
votes
3 answers

What kind of road and weather conditions can a self-driving car deal with?

Can self-driving cars deal with snow, heavy rain, or other weather conditions like these? Can they deal with unusual events, such as ducks on the road?
liori
  • 513
  • 2
  • 9
7
votes
2 answers

What was the first machine that was able to carry on a conversation?

What was the first AI that was able to carry on a conversation, with real responses, such as in the famous 'I am not a robot. I am a unicorn' case? A 'real response' constitutes a sort-of personalized answer to a specific input by a user.
Mithical
  • 2,905
  • 5
  • 27
  • 39
7
votes
1 answer

How should the neural network deal with unexpected inputs?

I recently wrote an application using a deep learning model designed to classify inputs. There are plenty of examples of this using images of irises, cats, and other objects. If I trained a data model to identify and classify different types of…
Jeff
  • 173
  • 5
7
votes
2 answers

How to calculate the number of parameters of a convolutional layer?

I was recently asked at an interview to calculate the number of parameters for a convolutional layer. I am deeply ashamed to admit I didn't know how to do that, even though I've been working and using CNN for years now. Given a convolutional layer…
Ælex
  • 205
  • 1
  • 2
  • 7
7
votes
2 answers

Interpretation of inverse matrix in mean calculation in Gaussian Process

The formula for mean prediction using Gaussian Process is $k(x_*, x)k(x, x)^{-1}y$, where $k$ is the covariance function. See e.g. equation 2.23 (in chapter 2) from Gaussian Processes for Machine Learning (2006) by C. E. Rasmussen & C. K. I.…
Kasia
  • 303
  • 2
  • 8
7
votes
2 answers

Why are reinforcement learning methods sample inefficient?

Reinforcement learning methods are considered to be extremely sample inefficient. For example, in a recent DeepMind paper by Hessel et al., they showed that in order to reach human-level performance on an Atari game running at 60 frames per second…
rrz0
  • 263
  • 2
  • 7
7
votes
2 answers

Does a quantum computer resolve the halting problem and would that advance strong AI?

Have there been proposed extensions to go beyond a Turing machine that solve the halting problem and if so, would those proposed extensions have value to advance strong Artificial Intelligence? For example, does quantum computing go beyond the…
WilliamKF
  • 2,513
  • 1
  • 25
  • 31
7
votes
1 answer

Medical diagnosis systems based on artificial neural networks

Are there any medical diagnosis systems that are already used somewhere that are based on artificial neural networks?
vojtak
  • 173
  • 2
7
votes
2 answers

What is the difference between artificial intelligence and cognitive science?

Sometimes I understand that people doing cognitive science try to avoid the term artificial intelligence. The feeling I get is that there is a need to put some distance to the GOFAI. Another impression that I get is that cognitive science is more…
Luis
  • 538
  • 7
  • 15
7
votes
1 answer

Why is the state-action value function used more than the state value function?

In reinforcement learning, the state-action value function seems to be used more than the state value function. Why is it so?
Bhuwan Bhatt
  • 394
  • 1
  • 12
7
votes
2 answers

Should I prefer the model with the lowest validation loss or the highest validation accuracy to deploy?

I trained a ResNet20 on Cifar10 and obtained the following learning curves. From the figures, I see at epoch 52, my validation loss is 0.323 (the lowest), and my validation accuracy is 89.7%. On the other hand, at the end of the training (epoch…
SpiderRico
  • 990
  • 9
  • 18
7
votes
2 answers

How can the theory of multiple intelligences be incorporated into AI?

I have been wondering since a while ago about the theory of multiple intelligences and how they could fit in the field of Artificial Intelligence as a whole. We hear from time to time about Leonardo Da Vinci being a genius or Bach's musical…
Luis
  • 538
  • 7
  • 15
7
votes
2 answers

How to estimate the capacity of a neural network?

Is it possible to estimate the capacity of a neural network model? If so, what are the techniques involved?
jaeger6
  • 308
  • 1
  • 7
7
votes
4 answers

Are PAC learnability and the No Free Lunch theorem contradictory?

I am reading the Understanding Machine Learning book by Shalev-Shwartz and Ben-David and based on the definitions of PAC learnability and No Free Lunch Theorem, and my understanding of them it seems like they contradict themselves. I know this is…
7
votes
1 answer

Why do ResNets avoid the vanishing gradient problem?

I read that, if we use the sigmoid or hyperbolic tangent activation functions in deep neural networks, we can have some problems with the vanishing of the gradient, and this is visible by the shapes of the derivative of these functions. ReLU solves…
FraMan
  • 199
  • 1
  • 3
  • 10