Dear Nandita Chaudhuri , you might find this paper that suggests an LSTM approach to predict creativity interesting:

https://link.springer.com/content/pdf/10.1007/s11571-020-09569-7.pdf

or this more general overview:

Preprint Creativity and Machine Learning: a Survey

Nandita Chaudhuri ,

How to mathematically model creativity?

Depends on what aspect do you want to model. To make it tractable, you would need to break it down to either the inputs, how are the inputs processed or on the outputs. You also need to narrow down the type of creative activity that you want to measure. It is not the same to measure painting creativity as opposed to technological artefact design. This scope narrowing will help you define the individual parameters that you want to capture for the model

Can we use deep learning models to predict creativity?

The answer is maybe. You cannot predict something that is just a vague concept. While you can throw any type of data to ML, it has to be a meaningful input to produce a relevant output. Doing the model first will help you narrow your query and refine the question so that you can design a good data gathering for your algorithm experiment.

Creativity, imagination and feelings is still a wide area of research in artificial intelligence. It's not yet Modeled in any artificial intelligence or deep learning based models.

The question here is: how do you define creativity?

In its literal meaning creativity is relative to each field and/or person and cannot be numerically expressed as far as I understand. If you define it as a combination of a set of metrics you can monitor, then it could be possible, but it would stil be a relative to your perspective/task/definition and its generalization be far from possible.

Dear Nandita Chaudhuri

Please take a look at the following paper, might be helpful:

Conference Paper Creative Generation of 3D Objects with Deep Learning and Inn...

Kind regards,

Cecilia-Irene Loeza-Mejía

Also, please check this creativity dataset, you could apply some deep learning algorithm

Article Multimodal brain data and core dimensions of creativity

Two interesting questions. The first needs to be addressed on multiple levels. On the micro level, several companies including Intel and IBM have been developing "neuromorphic processors." The Wikipedia article on neuromorphic engineering provides an interesting equation (called the "Caravelli-Traversa-Di Ventra equation") for the memory properties of components of such circuits. Once this is applied across the million (today) "neurons" for a chip, interesting outcomes occur. This approach is still in its infancy, but the potential for self-learning chips has already been demonstrated. How these might scale up to the macro level is a key to answering your first question. Nevertheless, the amazing superiority of neuromorphic processing over deep learning for some intractable tasks implies that the deep learning approach may not get to creativity, but new types of processors might have a better chance.

Some help on the macro level might come from my own research on the development of human practices. In Teachers, learners, modes of practice: Theory and methodology for identifying knowledge development" (2017) I provide a generalization of Lotka-Volterra competition that I call the "succession model. " It fits large (approximately 1,000) sample data on developmental transformations within single dimensions of practice including drawing, writing, and psychological research methods.

I also provide evidence there that integration across multiple dimensions is a prime source of creativity. Key to this argument is the finding that 80 design experts that interviewed identified 20 dimensions of development each of which contained three transformations to progressively more complex modes of practice. Twenty dimensions of 4 modes of practice each gives a trillion pattern. Generating novelty, therefore, is fairly trivial. To be creative, a novel result must also be useful. Since modes of practice compete with each other (in a Lotka-Volterra way) for use by humans, survival guarantees usefulness. Whether such survival is good for humanity will be an undecided question for quite a few decades.

Thank you Prof. David Dirlam for this insightful response.

Dear Dr Nandita Chaudhuri . See the following useful RG link: Conference Paper Predicting mathematical creativity

Kindly check also the following useful RG link: Article Mathematical Creativity: A Combination of Domain-general Cre...

Thank you Dr. David Dirlam , Dr. Arturo Geigel , Dr. Omar Amri

You are most welcome dear

Dr Nandita Chaudhuri

Wish you the best always

You are most welcome Dr Nandita Chaudhuri

Thank you Nandita Chaudhuri .