I am a pediatric cardiologist and have cared for children with heart disease for the past three decades. In addition, I have an educational background in business and finance as well as healthcare administration and global health – I gained a Masters Degree in Public Health from UCLA and taught Global Health there after I completed the program.
“Unity makes strength.”
I had the distinct honor of giving an opening keynote address for the Mayo Clinic Artificial Intelligence in Cardiology meeting last week (thank you Dr. Francisco Lopez-Jimenez).
I was asked by Dr. Lopez-Jimenez to speak on the burgeoning Alliance of Centers of AI in Medicine (ACAIM) and its pediatric congener the Pediatric Centers of AI in Medicine (PCAIM) and its impact on the future of artificial intelligence in the field of cardiology.
I took this unique opportunity to extend the topic to collaborative learning in all its forms (machine-to-machine or model-to-model, machine-to-human, and human-to-human) for all of us in this nascent field of artificial intelligence in medicine.
I think we are all inspired by the biological examples of collaborative or collective intelligence in swarms of animals such as fish, bees, and ants. One can perhaps even think of our individual brains as a collaborative network of neurons (about 100 billion) and synaptic connections (about 100 trillion). Let us imagine the collaborative learning potential of all of our brains and all of our machines as well.
I first discussed examples of machine-to-machine or model-to-model collaborative learning in the form of meta models used in machine learning ensemble methods such as bagging (parallel), boosting (sequential), and stacking. These efforts are designed to combine relatively weak learners so that the aggregate of these models is superior to the individual ones.
Another example of this model-to-model collaboration is generative adversarial network (GAN) as this deep neural network framework is in reality two neural networks in the form of a generator and a discriminator that “compete” against each other so that the generator learns to produce more examples that are plausible.
A third possible example of this model-to-model collaborative learning is transfer learning as this concept involves deep neural networks facilitating learning from a previous model that trained from a prior data set. In other words, the knowledge from a prior model can be transferred to another model to facilitate its solving of a related problem.
In this coming decade and next, we will see more and more examples of models and collaborative learning in its many permutations as artificial intelligence evolves beyond machine and deep learning into a more cognitive era.
We will continue this theme of collaborative learning in machine-to-human and human-to-human contexts in the next newsletter, which will be every two weeks throughout August.
We are looking forward to our next next opportunity to have collaborative learning at the AIMed Global Summit 2023 scheduled for Spring of 2023 in southern California.
Hope to see you many times before then!