Learning & GenAI
Learning & Memory
-
Can machine learning, particularly GenAI, help in learning? Where is the connection? READ MORE HERE
Machine Learning
There are a lot of theories for explaining learning by humans and the involvement of biological neural network. Artificial Intelligence (AI), the term which was originally coined at Dartmouth College in 1956, is based on attempting to mimicking this neural network by developing machine learning algorithms, particularly Deep learning algorithm. Among these, the newly developed 'Transformer' model by Vaswani et al (2017) is holding the promise of developing this machine learning algorithm better. A new report explains how the brain adjusts connections between neurons during learning. This new insight may inspire developing faster and more robust learning algorithms in artificial intelligence.
GenAI, such as Gemini, interestingly 'learn' more and more from its use and the quality of this process largely depends on the proper 'prompting'. Google, the developer of this 'Transformer' model and the pioneer of search engine development, recently released Prompting Guide 101 for the end users. Ultimate goal is to 'train' their Gemini for performing better from the 'feedback' of the end users. At the back end, both the domain experts and the developers can then fine tune the algorithm for better performance and reducing the 'hallucination'.
Our Advisors Our Collaborators