Daniel Crespin

13 Questions 70 Answers 0 Followers

Questions related from Daniel Crespin

Is backpropagation really used in practical, commercially oriented AI applications?

Hello to all Artificial intelligence is often mentioned as an area where corporations make large investments. It seems that they use AI in autonomous vehicles, face recognition, medicine,...

10 October 2018 2,865 6 View

Assume that Quantum Mechanics is incomplete and mistaken, and the correct replacement is found, is not better for the QM establishment to ignore it?

The question and this explanation were posted elsewhere, but initial answers disappeared from the site. I repeat it here hoping it will stay. %%%%%%%% There are several Nobel Physics winners that...

06 June 2018 6,529 5 View

Are random quantum jumps the only possibility?

Schrödinger self adjoint operator H is crucial for the current quantum model of the hydrogen atom. It essentially specifies the stationary states and energies. Then there is Schrödinger unitary...

11 November 2016 4,160 62 View

Have you seen this group?

Let R^* be the multiplicative group of non-zero real numbers, R the additive group of real numbers, and consider on the set R^*\times R the group operation defined as (a,b)*(a',b') = (aa',b+ab') I...

04 April 2016 1,026 4 View

Is quantum entanglement any better than classical entanglement?

Reading the opening post of a software...

04 April 2014 8,740 6 View

What does "breakthrough in artificial intelligence, so machines can learn" mean?

Bill Gates made the following statement "If you invent a breakthrough in artificial intelligence, so machines can learn, that is worth 10 Microsofts." (The New York Times, March 3, 2004) Mr....

03 March 2014 878 4 View

For neural network training, what is better than backpropagation?

The usual way to obtain neural networks that recognize given data is incremental training by backpropagation. But there are much better alternatives. The following is taken from section 3 of the...

04 April 2013 5,493 9 View

After a neural network is trained. Can it really be validated?

Does it make sense to validate a NN after training? Let A, B be training sets of n-vectors. Vectors in A are examples, or vectors with expected output 1. Vectors in B are counterexamples, with...

04 April 2012 7,577 20 View

What is the status of the n-dimensional XOR training problem for neural networks ?

The XOR problem in dimension 2 appears in most introductory books on neural networks. The problem consists four 2-dimensional binary data vectors, each with a binary output 0 or 1 according to the...

03 March 2012 9,061 9 View

Where to find numerical solutions of the Sonar, or Mines vs Rocks, neural network training problem?

Several online repositories exist that provide numerical data for machine learning problems. One of these is http://archive.ics.uci.edu/ml/index.html There you can find the Sonar, or Mines vs...

03 March 2012 6,116 1 View

Assume a way is found to instantly obtain the architecture and weights of multilayer neural networks that perfectly recognize given data. Will it make...

Under the above assumption ALL data can be used for training. Testing is useless since the NN obtained is known to correctly classify all the data. Also, if new data are provided that the old NN...

02 February 2012 1,535 2 View

How to start marketing a NEW NEURAL NETWORK TECHNOLOGY?

Assume that a new neural network procedure (NNNP) is found that allows to efficiently train neural networks up to the best their theoretical performance. Suppose that the NNNP is "miraculous" and...

01 January 1970 7,716 7 View

Determinism vs Entanglement. Why call it ENTANGLEMENT when it actually is plain DETERMINISM?

Physical experiments were performed by Alain Aspect in 1982, later modified and continued by others. In the experiments results of far separated measurements of certain particle observables were...

01 January 1970 7,199 11 View