Do you think that AI will ever feel emotions?

[Steve Klinko]

This is another on of those baseless, blue-sky claims.

Show that "Designers use the Pattern Recognition capability and nothing else."

How do you expect to be taken seriously making such a claim?

I think this was the line by billvon that started this:
Figure out the most efficient trajectories. Learn to fly airplanes better. Understand language. Correct grammar. Predict weather. Paint pictures. Write stories. When installed in a robot, figure out how to walk and run.

What are the Artificial Neural Nets doing in each of these tasks besides, Pattern Matching within the dedicated Task Algorithm? It takes a lot of other kinds of Software to accomplish these tasks. More than what can be accomplished with just an ANN.

 

[DaveC426913]

I think this was the line by billvon that started this:
It takes a lot of other kinds of Software to accomplish these tasks. More than what can be accomplished with just an ANN.

Again: says you. Without anything to back it up.

You are using your own beliefs of what AI can and can't do to try to support an argument about what AI can and can't do.

 

[billvon]

When systems are designed using Artificial Neural Nets the Designers are using the Pattern Recognition capability and nothing else.

Not true. Google deep learning, for example. You're about 15 years behind the times.

 

[Steve Klinko]

Again: says you. Without anything to back it up.

You are using your own beliefs of what AI can and can't do to try to support an argument about what AI can and can't do.

As an Engineer, I work on AI projects from time to time, and I can say with 100% certainty that you are trying to stuff capability into Neural Nets beyond what they are really doing.

 

[Steve Klinko]

Not true. Google deep learning, for example. You're about 15 years behind the times.

I was very disappointed when I studied up on Deep Learning a bunch of years ago. Nothing new just extra layers in the Net. It provided better Pattern Recognition and faster configuration times. No Magic new Category of capability from previous Neural Nets. Deep Learning is Marketing Hype for AI Software companies that want to sell products.

 

[Write4U]

What are the Artificial Neural Nets doing in each of these tasks besides, Pattern Matching within the dedicated Task Algorithm? It takes a lot of other kinds of Software to accomplish these tasks. More than what can be accomplished with just an ANN.

Within 10 minutes of its birth, a baby fawn is able to stand. Within seven hours, it is able to walk. Between those two milestones, it engages in a highly adorable, highly frenetic flailing of limbs to figure it all out.

The procedure used to carry out the learning process in a neural network is called the optimization algorithm (or optimizer).

There are many different optimization algorithms. All have different characteristics and performance in terms of memory requirements, processing speed, and numerical precision.

In this post, we formulate the learning problem for neural networks. Then, some important optimization algorithms are described. Finally, the memory, speed, and precision of those algorithms are compared.

• Learning problem.
• 1. Gradient descent.
• 2. Newton method.
• 3. Conjugate gradient.
• 4. Quasi-Newton method.
• 5. Levenberg-Marquardt algorithm.
• Performance comparison.
• Conclusions.

Learning problem

The learning problem is formulated in terms of the minimization of a loss index, f" role="presentation" style="display: inline-block; line-height: 0; font-size: 19.52px; letter-spacing: normal; overflow-wrap: normal; word-spacing: normal; white-space: nowrap; float: none; direction: ltr; max-width: none; max-height: none; min-width: 0px; min-height: 0px; border: 0px; margin: 0px; padding: 1px 0px; position: relative;">ff. It is a function that measures the performance of a neural network on a data set.

The loss index is, in general, composed of an error and a regularization terms. The error term evaluates how a neural network fits the data set. The regularization term is used to prevent overfitting by controlling the sufficient complexity of the neural network. .......more

https://www.neuraldesigner.com/blog/5_algorithms_to_train_a_neural_network#:

A Detailed Guide to 7 Loss Functions for Machine Learning Algorithms with Python Code

Loss functions are at the heart of the machine learning algorithms we love to use. But I’ve seen the majority of beginners and enthusiasts become quite confused regarding how and where to use them.

They’re not difficult to understand and will enhance your understand of machine learning algorithms infinitely. So, what are loss functions and how can you grasp their meaning?.......more

https://www.analyticsvidhya.com/blo...-loss-functions-machine-learning-python-code/

Neural Network Machine Learning Algorithms

Machine learning algorithms that use neural networks typically do not need to be programmed with specific rules that outline what to expect from the input.

Perceptron

A neural network is an interconnected system of the perceptron, so it is safe to say perception is the foundation of any neural network. It is a binary algorithm used for learning the threshold function.....more

https://www.educba.com/neural-network-machine-learning/

 

[billvon]

I was very disappointed when I studied up on Deep Learning a bunch of years ago.

Well there's your problem. Your 15 year old knowledge is no longer valid.

 

[James R]

. It will have to be pretty obvious, like all of a sudden seeing circles, squares or spiral patterns.

How will you tell that those aren't random?

 

[Steve Klinko]

Within 10 minutes of its birth, a baby fawn is able to stand. Within seven hours, it is able to walk. Between those two milestones, it engages in a highly adorable, highly frenetic flailing of limbs to figure it all out.

• Learning problem.
• 1. Gradient descent.
• 2. Newton method.
• 3. Conjugate gradient.
• 4. Quasi-Newton method.
• 5. Levenberg-Marquardt algorithm.
• Performance comparison.
• Conclusions.

Learning problem

https://www.neuraldesigner.com/blog/5_algorithms_to_train_a_neural_network#:

A Detailed Guide to 7 Loss Functions for Machine Learning Algorithms with Python Code

https://www.analyticsvidhya.com/blo...-loss-functions-machine-learning-python-code/

Neural Network Machine Learning Algorithms
Perceptron https://www.educba.com/neural-network-machine-learning/

It's all ways of configuring the Neural Net. When the Net is configured it's all Pattern Matching. The Perceptron was conceived back in the 50s. Your information is 70 years old.

 

[Steve Klinko]

How will you tell that those aren't random?

Could be random, but after watching actual random patterns for a long time, it will be quite apparent that something is going on if the patterns start consistently organizing like this.

 

[Steve Klinko]

Well there's your problem. Your 15 year old knowledge is no longer valid.

Where did the 15 years come from?

 

[Write4U]

It's all ways of configuring the Neural Net. When the Net is configured it's all Pattern Matching. The Perceptron was conceived back in the 50s. Your information is 70 years old.

So, is the concept useless or has it been refined? Seems to me most scientific theories have stood the test of time.

 

[billvon]

Where did the 15 years come from?

That's about the time that deep learning started to become a serious field of research.

 

[Write4U]

This is new.

MuZero

MuZero is a computer program developed by artificial intelligence research company DeepMind to master games without knowing their rules.[2][3][4] Its release in 2019 included benchmarks of its performance in go, chess, shogi, and a standard suite of Atari games. The algorithm uses an approach similar to AlphaZero. It matched AlphaZero's performance in chess and shogi, improved on its performance in Go (setting a new world record), and improved on the state of the art in mastering a suite of 57 Atari games (the Arcade Learning Environment), a visually-complex domain.

MuZero was trained via self-play and play against AlphaZero, with no access to rules, opening books, or endgame tables. The trained algorithm used the same convolutional and residual algorithms as AlphaZero, but with 20% fewer computation steps per node in the search tree.[5]

https://en.wikipedia.org/wiki/MuZero

 

[Steve Klinko]

That's about the time that deep learning started to become a serious field of research.

It was hiding in the basement of Academia 10 years before I came onboard. I started looking into it when I started hearing the Marketing Adds for AI promoting Deep Learning. Seems like the Marketing adds finally hit my Consciousness about 5 years ago.

 

[Steve Klinko]

So, is the concept useless or has it been refined? Seems to me most scientific theories have stood the test of time.

Of course it was refined. But a Neural Net or Perceptron or whatever you call it, is still a Pattern Matching tool.

 

[Steve Klinko]

This is new.

MuZero https://en.wikipedia.org/wiki/MuZero

Scientists realized that Games were very Algorithmic and Pattern Driven many years ago. Self configuring Neural Nets are not difficult to program, and this has been done for many years. I think the Company in your link has a good product and a good Marketing strategy.

 

[Write4U]

Scientists realized that Games were very Algorithmic and Pattern Driven many years ago. Self configuring Neural Nets are not difficult to program, and this has been done for many years.

And what is the difference? Brains are not pattern matching processors? I told you to watch Anil Seth.

I think the Company in your link has a good product and a good Marketing strategy

And you believed these people were not seriously trying to create a functional product?
AI is the future, my friend. We better get on board with that.

 

[Steve Klinko]

And what is the difference? Brains are not pattern matching processors? I told you to watch Anil Seth.
And you believed these people were not seriously trying to create a functional product?
AI is the future, my friend. We better get on board with that.

You are comparing Brain Neural Nets with Artificial Neural Nets. Both do Pattern Matching but the Brain does much more. You are comparing Apples with Bowling Balls again.

 

[Write4U]

You are comparing Brain Neural Nets with Artificial Neural Nets. Both do Pattern Matching but the Brain does much more. You are comparing Apples with Bowling Balls again.

What do you think a brain is ?
As Descartes proved, if you place a brain in a vat attached to a computer and play an outdoor scene, the brain will think it is outdoors walking in the sun. Yes it will imagine more than just experiencing outdoors, but it'll still be in the vat.

Now suppose you do this with an AI and in addition to an image of the sun you also feed in warmth. Will the AI's thermometer sense the warmth and seeing the sun, imagine a sunlit scene from memory?