30 votes

How to select number of hidden layers and number of memory cells in an LSTM?

Your question is quite broad, but here are some tips. Specifically for LSTMs, see this Reddit discussion Does the number of layers in an LSTM network affect its ability to remember long patterns? ...
user avatar
29 votes
Accepted

What are some well-known problems where neural networks don't do very well?

Here's a snippet from an article by Gary Marcus In particular, they showed that standard deep learning nets often fall apart when confronted with common stimuli rotated in three dimensional ...
user avatar
28 votes

What are the current theories on the development of a conscious AI?

To answer this question, first we need to know why developing conscious AI is hard. The main reason is that there is no mathematically or otherwise rigorous definition of consciousness. Sure you ...
user avatar
28 votes
Accepted

Is there any research on the development of attacks against artificial intelligence systems?

Yes, there is some research on this topic, which can be called adversarial machine learning, which is more an experimental field. An adversarial example is an input similar to the ones used to train ...
user avatar
  • 33.8k
26 votes
Accepted

Which explainable artificial intelligence techniques are there?

Explainable AI and model interpretability are hyper-active and hyper-hot areas of current research (think of holy grail, or something), which have been brought forward lately not least due to the (...
user avatar
20 votes
Accepted

Where can I find the proof of the universal approximation theorem?

There are multiple papers on the topic because there have been multiple attempts to prove that neural networks are universal (i.e. they can approximate any continuous function) from slightly different ...
user avatar
  • 33.8k
19 votes

What are some well-known problems where neural networks don't do very well?

In theory, most neural networks can approximate any continuous function on compact subsets of $\mathbb{R}^n$, provided that the activation functions satisfy certain mild conditions. This is known as ...
user avatar
  • 33.8k
16 votes

Are there other approaches to deal with variable action spaces?

Does anyone know any paper regarding this subject? I'm not familiar with any off the top of my head. I do know that the vast majority of Reinforcement Learning literature focuses on settings with a ...
user avatar
  • 9,369
16 votes

What are examples of promising AI/ML techniques that are computationally intractable?

AIXI is a Bayesian, non-Markov, reinforcement learning and artificial general intelligence agent that is incomputable, given the involved incomputable Kolmogorov complexity. However, there are ...
user avatar
  • 33.8k
16 votes

How do neural networks play chess?

Minimax and related algorithms are used to play chess. That is how chess programs have worked for many years (with some additions such as standard opening playbooks). They do not need to process the ...
user avatar
  • 23.8k
15 votes
Accepted

When should I use 3D convolutions?

3D convolutions are used when you want to extract features in 3 dimensions or establish a relationship between 3 dimensions. Essentially, it's the same as 2D convolutions, but the kernel movement is ...
user avatar
  • 1,359
15 votes

What are some well-known problems where neural networks don't do very well?

In our deep learning lecture, we discussed the following example (from Unmasking Clever Hans predictors and assessing what machines really learn (2019) by Lapuschkin et al.). Here the neural network ...
user avatar
  • 251
13 votes
Accepted

Are the dialogs at Sophia's (the robot) appearings scripted?

Sophia uses ChatScript. You can read about what ChatScript can do here. ChatScript keeps track of conversations with each user; can record where it is in a conversational flow and what facts it ...
user avatar
13 votes

What are examples of promising AI/ML techniques that are computationally intractable?

Exact Bayesian inference is (often) intractable (i.e. there is no closed-form solution, or numerical approximations are also computationally expensive) because it involves the computation of an ...
user avatar
  • 33.8k
12 votes

How could we build a neural network that is invariant to permutations of the inputs?

Here is a few that might be what you are looking for: Deep Sets, https://papers.nips.cc/paper/6931-deep-sets.pdf BRUNO: A Deep Recurrent Model for Exchangeable Data, https://arxiv.org/pdf/1802.07535....
user avatar
  • 221
12 votes

Is there any research on the development of attacks against artificial intelligence systems?

Sometimes if the rules used by an AI to identify characters are discovered, and if the rules used by a human being to identify the same characters are different, it is possible to design characters ...
user avatar
  • 353
11 votes
Accepted

How can I automate the choice of the architecture of a neural network for an arbitrary problem?

I think in this case, you'll probably want to use a genetic algorithm to generate a topology rather than working on your own. I personally like NEAT (NeuroEvolution of Augmenting Topologies). The ...
user avatar
  • 226
11 votes

Where can I find the original paper that introduced RNNs?

The two tech reports below both call RNNs explicitly "recurrent net(work)s". Rumelhart, David E; Hinton, Geoffrey E, and Williams, Ronald J (Sept. 1985). Learning internal representations ...
user avatar
11 votes

Is there any research on the development of attacks against artificial intelligence systems?

Yes there are, for instance one pixel attacks described in Su, J.; Vargas, D.V.; Kouichi, S. One pixel attack for fooling deep neural networks. arXiv:1710.08864 One pixels attacks are attacks in ...
user avatar
10 votes
Accepted

Can an AI make a constructed (natural) language?

could an AI make/construct its own natural language, with words, conjugations, and grammar rules? Sure. This might be helpful: Simulated Evolution of Language: a Review of the Field Basically, a ...
user avatar
  • 1,837
10 votes
Accepted

How could we build a neural network that is invariant to permutations of the inputs?

Traditionally, due to the way the network is structured, each input has a set of weights, that are connected to more inputs. If the inputs switch, the output will too. Approach 1 However, you can ...
user avatar
10 votes

What are the mathematical prerequisites for an AI researcher?

I work as a professor, and recently designed the mathematics requirements for a new AI major, in consultation with many of my colleagues at other institutions. The other answers, particularly this one ...
user avatar
10 votes

What are examples of promising AI/ML techniques that are computationally intractable?

This question gets at a really interesting fact about AI research in general: AI is hard. In fact, almost every AI problem is computationally hard (typically NP-Hard, or #P-Hard). This means that ...
user avatar
10 votes

What are examples of approaches to create an AI for a fighting robot in an MMO game?

I would set up a list of goals for your bot. These could be 'maintain a minimum level of health', 'knock out human player', 'block way to location X', etc. This obviously depends on the domain of your ...
user avatar
  • 5,062
10 votes
Accepted

What are some online courses on artificial general intelligence?

As far as I know, no AGI system has yet been created, so that's why there aren't yet many courses on AGI. However, there are a few courses that attempt to address AGI as the main topic but from ...
user avatar
  • 33.8k
10 votes

How do neural networks play chess?

This is a good question. your understanding in general is correct. Indeed, data can be used to construct a proper evaluation of a move/board position and recommended moves based on its history (at ...
user avatar
  • 739
9 votes

Is there any board game where a human can still beat an AI?

Not all games (or even board games) are computationally algorithmic. Even the least skilled player is likely to trounce the hottest pattern-matching algorithm in a game of Pictionary (for example). ...
user avatar
9 votes

Can AI write good jokes yet?

I don't think the AI has gotten to that point yet. Here are some of the interesting papers on the subject: A paper was recently written that attempted to generate jokes using unsupervised learning. ...
user avatar
8 votes
Accepted

Is there any artificial intelligence that possesses "concentration"?

Douglas Hofstadter's CopyCat architecture for solving letter-string analogy problems was deliberately engineered to maintain a semantically-informed notion of 'salience', i.e. given a variety of ...
user avatar

Only top scored, non community-wiki answers of a minimum length are eligible