In this episode, I will outline my concept on “How to reach the King’s Scholarship level?”. This is where my understanding of machine learning and dynamic simulation will help.

My approach is focused on motivation because I believe that if the child does not want it, then no amount of parental push or even Asian parents would make the process sustainable or enjoyable. So step 1 would be to gauge the interest in learning. Step 2 to light a passion, and the motivation and mindset will come along. The probability of completing the King’s Scholarship exams confidently (hopefully correctly) and excelling in interviews and interesting personal statements will increase. These King Scholarships are not impossible, just improbable due to limited places.

There are some assumptions first.

1. Every child is different. – I will tailor the approach to each individual while deriving a systematic way to achieve the goal.
2. Every child has great potential in their own way. – Not everyone is made to be a mathematician. Everyone has their own spark. The joy is to reach the best level
3. Learning is enjoyable and fun. – I enjoyed researching and learning for my PhD. Stay humble and curious, and keep looking for knowledge.

The Science of Learning, such as cognitive science, inspired Machine Learning. Although Kids are not precisely machines, I see commonalities which may help. For example, the multi-layers neural network in deep learning can be similar to a child’s mind, ready to process a massive amount of information. The unsupervised learning approach to train a deep neural network results in the machine being able to do the right thing, like autonomous driving, but there is unpredictability. As each layer of abstraction is unclear. So we are not sure how deep neural network come up with the right answer. Does it matter? We should encourage a child can find their way of solving a problem, if it works, it works. Unsupervised learning will be one of the key ingredients.

Supervised learning is the opposite of unsupervised learning. Another person can supervise and decipher the child’s abstraction of the concepts to make sense or not. The key danger is over-fitting in machine learning or over-tuning a physical dynamic simulation model. The neural network or model will only work for a given set of conditions and parameters. The child can only resolve a given problem in a particular format. A balance between understanding the child’s abstraction of a topic and telling them how things work needs to be delicate. The causal approach of explaining a subject to a child may impact their overall understanding of the subject in terms of defining a rigid understanding of their abstractions, which may limit creativity. In a neural network, a bias, which is detrimental to other useful neural pathways. There is always a place for supervised learning to keep the abstraction relevant. and able to find the solutions.

In conclusion, a balance between unsupervised learning for the child to build their own abstractions on the subject and supervised learning to verify the abstractions will reach the KS level. An acausal approach to learning will draw the big-picture understanding that will be critical in solving these exam questions. I will provide guidance and “tuition” to enable a child, a little learning machine, to achieve a high academic level and complete these King’s Scholarship exams to a high standard. To realise my aim that King’s Scholarship level is not limited to certain schools or backgrounds.