Randy
  • Member for 5 years, 5 months
  • Last seen this week
  • Outer Philadelphia, PA
When is deep learning overkill?
Accepted answer
13 votes

It's all about Return On Investment. If DL is "worth doing", it's not overkill. If the cost of using DL (computer cycles, storage, training time) is acceptable, and the data available to train it is ...

View answer
Could we code in Natural language once we pass the Turing test?
1 votes

No. Humans pass the Turing Test but cannot express themselves with enough precision and little enough ambiguity to code well in English (natural language). No machine will improve on that until it ...

View answer
Is it possible to train a neural network to estimate a vehicle's length?
0 votes

Yes it's possible, but first you'll have to recognize some object in the image, either 1) the vehicle itself, and then report that vehicle's known size, or 2) a known object that's the same distance ...

View answer
Are probabilistic models dead ends in AI?
2 votes

I think Minsky deprecated the suggestion that probabilistic models could be surrogates for component models for intelligence that he suggested were grounded in principles and processes that interact (...

View answer
Precise localization and characterization of rudimentary shapes with neural networks
2 votes

I think you're describing "object localization and detection" which combines object identification with discovery of its spatial placement in the field of view. There's been a lot of work on this in ...

View answer
Is there a formal name for this philosophical AI problem?
1 votes

Your problem closely resembles John Searle's "Chinese Room" argument, which claimed that one (or more) abstract "intelligence tests" lack the discriminitive ability to distinguish between a trivial ...

View answer
Why are inhibitory connections often used in virtual neural networks when they don't seem to exist in real life neural networks?
2 votes

From what I've read on deep nets and neuroscience, your interpretation is backwards. Most 'neurons' in deep nets use RELU activations, which is strictly excitatory, while mammal brains are composed ...

View answer
Are Convolutional Neural Networks better than existing image recognition libraries that don't use CNNs?
0 votes

There are object recognition tasks where DL-CNNs are not yet state of the art, like pedestrian detection. Probably this is because the task is considerably more complex than simple visual object ...

View answer
Different useful approaches of implementing real-time AI?
1 votes

About 15 years ago, John Laird's group at Michigan used the Soar rule-based architecture to play several FPS games effectively (Quake II, Descent III): http://ai.eecs.umich.edu/people/laird/...

View answer
What does Brooks mean by "representation"?
1 votes

In that paper, Brooks introduced the basis for what became known as his "subsumption architecture". The idea was to get away from the 1980's popular approach of a single global representation of all ...

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

No, smart cars do not know what to do when surrounded with ducks or flood waters, and it's possible they never will. As with all machine learning, a computer knows only what it's taught. If an ...

View answer
Would an AI with human intelligence have the same rights as a human under current legal frameworks?
Accepted answer
6 votes

Yes, to some of what you propose. No to some. Today corporations are granted rights: to own property, earn income, pay taxes, contribute to political campaigns, offer opinion in public, ad more. ...

View answer
Is Artificial Intelligence restricted to electrical based technology?
5 votes

No, I think electricity is not essential for AI. In theory AI (a sufficient collection of computational processes that can adapt to changes in their input, thus producing 'intelligent' behavior), ...

View answer
Is AI programming useful in everyday programs?
2 votes

I believe AI is rarely used in mainstream apps, but it could be, and I think slowly will be. If the information an app's AI must learn arises within the app, from user interaction or error, it'd be ...

View answer
What jobs cannot be automatized by AI in the future?
1 votes

Probably the only secure jobs are those where the audience enjoys watching live human craftsmanship take place in real time right before their eyes, like acting or standup comedy or musical virtuosity ...

View answer
How can thought vectors be used outside of an Artificial Neural Network (ANN) context?
2 votes

I'll take a shot at answering this, though I'm no expert in Neural Nets or Deep Learning. Given that practical thought vectors (TVs) don't yet exist, and may be impractical or impossible, I think ...

View answer
How could self-driving cars make ethical decisions about who to kill?
4 votes

Frankly I think this issue (the Trolley Problem) is inherently overcomplicated, since the real world solution is likely to be pretty straightforward. Like a human driver, an AI driver will be ...

View answer
How does using ASIC for the acceleration of AI work?
3 votes

I think the algorithm has changed minimally, but the necessary hardware has been trimmed to the bone. The number of gate transitions are reduced (perhaps float ops and precision too), as are the ...

View answer
What purpose would be served by developing AI's that experience human-like emotions?
4 votes

I think emotions are not necessary for an AI agent to be useful. But I also think they could make the agent MUCH more pleasant to work with. If the bot you're talking with can read your emotions and ...

View answer
How close are we to creating Ex Machina?
1 votes

Based on the success of IBM Watson and the amazing advances in tackling numerous hard tasks using deep learning in the past 3 years, I think a large high-tech company like Google or Amazon will create ...

View answer