Hot answers tagged

42

Yes, indeed, neural networks are very prone to catastrophic forgetting (or interference). Currently, this problem is often ignored because neural networks are mainly trained offline (sometimes called batch training), where this problem does not often arise, and not online or incrementally, which is fundamental to the development of artificial general ...


16

Yes, the problem of forgetting older training examples is a characteristic of Neural Networks. I wouldn't call it a "flaw" though because it helps them be more adaptive and allows for interesting applications such as transfer learning (if a network remembered old training too well, fine tuning it to new data would be meaningless). In practice what you want ...


9

tl;dr I always like to think of Neural Networks as a generalization of logistic regression. I too don't like that, traditionally, when introducing Neural Networks, books start with biological neurons and synapses, etc. I think its more beneficial to start from statistics and linear regression, then logistic regression and then neural networks. A ...


6

Since a neural network does iteratively learn its own weights I assume you mean the structure of the neural network - the number of layers and nodes per layer. If what I said above was your question, then yes, it most definitely is being explored. Even when a neural network is allowed to learn its own structure it still needs to be suited to a specific ...


5

Interpreted languages allow for a faster development cycle, as they don't require time for compilation, and fragments can often be run without having a complete program. They often also have fewer constraints for variable declaration or typing. That means they can be used to quickly scope out a problem and try different solutions. The drawback is the slower ...


4

An important thing we're going to need is what is called the "Expected Grad-Log-Prob Lemma here" (proof included on that page), which says that (for any $t$): $$\mathbb{E}_{\tau \sim \pi_{\theta}(\tau)} \left[ \nabla_{\theta} \log \pi_{\theta}(a_t \mid s_t) \right] = 0.$$ Taking the analytical expression of the gradient (from, for example, slide 9) as a ...


4

What you are describing sounds like it could be a deliberate case of fine-tuning. There is a fundamental assumption that makes minibatch gradient descent work for learning problems: It is assumed that any batch or temporal window of consecutive batches forms a decent approximation of the true global gradient of the error function with respect to any ...


3

Yes, the notion is valid and it has been indeed explored to some extent, although we are still far away from a breath-taking result. These are the topics that have been explored in that regard and as far as I know: Regarding code generation, the most successful results imply the use of Domain-Specific Languages and/or strong restrictions. A neural policy ...


3

What you're describing is called a recurrent neural network. There are a large number of designs in this family that all have the ability to remember recent inputs and use them in the processing of future inputs. The "Long Short Term Memory" or LSTM architecture was one of the most successful in this family. These are actually very widely used in things ...


3

The acclaimed book Artificial Intelligence: A Modern Approach (by Stuart Russell and Peter Norvig) gives a definition of an agent An agent is anything that can be viewed as perceiving its environment through sensors and acting upon that environment through actuators. This definition is illustrated by the following figure This definition (and ...


3

That is a wonderfully fundamental question. Learning is the use of a system to change another system so that, instead of doing what it did before, which may have been nothing, it does something else. In the human brain, the system is the way that genetic expression caused the directed mutability of that brain so that human intentions and responses to ...


3

There is at least one very important and serious AI scientist that apparently believes in the creation of true artificial general intelligence and possibly superintelligence: Jürgen Schmidhuber, who is the co-author of the LSTM, among many other important contributions. In fact, he recently founded NNAISENSE for this ultimate purpose, that is, to build a ...


3

In this particular context, "Democratize" means to make more accessible to people. Thus, "Democratizing AI" means to make AI softwares and AI programming available, accessible and easy to use for the vast majority of people.


2

In addition to Jaden's excellent answer "no one is trying to actually make a “conscious” AI because we don’t know what that word means yet" I'd like to add that the word "yet" there is highly optimistic. It's highly problematic and likely impossible to distinguish between a conscious being and a being that behaves exactly as if it was conscious. ...


2

On the applicability of Artificial Intelligence in Black Box Testing (Khanna, 2017) might be a good place to start. It classifies some black box testing techniques based on the AI branch it belongs to.


2

I believe the question you're asking is whether or not it is possible for a neural network to edit it's own structure and learning ability intelligently. I would argue this is a matter of philosophy, of how intelligent can we program a computer to be, and I personally believe that yes, at some point humans will develop some form of an algorithm capable of ...


2

Can a neural network modify its own weights? One important step in training a neural network is called backpropagation. In the course of this process, the weights of the neural network are updated into a direction that minimizes the training loss. Usually, this step happens after each batch (with batch gradient descent) or after each sample (with stochastic ...


2

One way to view a neural network is as a series of linear transformations. You take a bunch of data points and look at it from a different perspective from a different space. You apply some non linear function on the data points like, ReLU, sigmoid etc. Now you repeat the same process of looking from a different space. Our goal is to look at it from a ...


2

When formulating a problem in deep learning, we need to come up with a loss function, which uses model weights as parameters. Back-propagation starts at an arbitrary point on the error manifold defined by the loss function and with every iteration intends to move closer to a point that minimises error value by updating the weights. Essentially for every ...


2

According to a nice article by Sebastian Ruder https://ruder.io/4-biggest-open-problems-in-nlp/ based on answers from top NLP researchers https://docs.google.com/document/d/18NoNdArdzDLJFQGBMVMsQ-iLOowP1XXDaSVRmYN0IyM/edit Natural language understanding NLP for low-resource scenarios Reasoning about large or multiple documents Datasets, problems, and ...


1

I am very surprised no one mentioned the 'Lottery Ticket Hypothesis' paper which won the best paper award in ICLR 2019. Although, the main idea has been presented as ways to reduce connections between successive nodes in a Neural Network, it can be viewed as a way Neural Network learning for itself which of its connections are not important. The main aim is ...


1

There are works on expandable neural networks, for example, Lifelong learning with dynamically expandable networks by Yoon et al. So, if you consider the whole system with expanding algorithm (in the same way as you consider an SGD optimizer as part of a network system), then yes, it's totally possible. But they didn't show breakthrough performance on any ...


1

While I dont know if something like what I will describe exist, I will give my idea of how this might be achieved. I think there are two ways, one being an nlp network that as you said, you "show the net its own code" and let if try to output code that runs but is different, likely this would be done in an rl environment with another problem domain to ...


1

There is absolutely zero evidence that an artificial neural network (ANN) can change itself. Arguably, self-modification requires some form of consciousness (that is, the state of being aware of and responsive to one's surroundings) or self-awareness (that is, conscious knowledge of one's own character and feelings), which (arguably) no ANN currently ...


1

Neural Networks aren't the same as human brains, thus as per that the first question you ask is maybe thinking too far. The neural networks cannot be aware of anything and thus not understand nor edit any code. OK, you further explained it would fine tune some parameters, but that I wouldn't call editing the code. Editing the algorithm to some other would ...


1

To answer your question I'd say: Maybe in theory, but not in practice. The problem is that you only consider a chronological/sequential training. Only once have I used such sequential training method that is called online-training or Online Machine Learning. That was using the woppal wabbit library. It is a feature (not an issue like you consider) of this ...


1

XAI is relevant to "black box" AI (machine learning methods where the decision making rationale is not apparent, only the structure of the system that led to that decision.) Symbolic AI, GOFAI, and Expert Systems are both explainable and understood, in that the the decision-making process is designed by humans. (Symbolic AI involves human-readable ...


1

A system is Turing complete if it can be used to simulate any Turing machine. Given the Church-Turing thesis (which has not yet been proven), a human brain can compute any function that a Turing machine can (given enough time and space), but the reverse is not necessarily true, given that the human brain might be able to compute more functions than a ...


1

My answer is yes, but in a trivial way. The least you would expect from an intelligent agent is that it is able to execute a given Turing machine on a given input. This requires actually no intelligence, just following rules. If however, you are referring to the capability of predicting if the Turing machine will terminate on the given input, that is another ...


1

Maybe, but it depends to a very large degree on the choice of definition. One of the biggest challenges for AI researchers, neuroscientists, philosophers, and psychologists, has been that the layperson's understanding of intelligence does not appear to correspond to a well-defined concept. This point was most famously exploited by John R. Searle in his ...


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