Human Learning (HL) vs. ML: Unique Techniques We Can Learn


Human Learning (HL) vs. Machine Learning (ML) has many unique features. If we master these unique features and techniques, we can not only improve our learning but also potentially use them for Machine Learning (ML). Here I would like to share a few useful HL techniques and compare them with ML. For instance, the diffused mode, sleep over the problem, the recall method, the chunking method, and etc.

In the context of AI and ML, we can compare the human learning techniques vs the machine learning techniques. We may find that these techniques are not widely used in today’s machine learning science at all. One of the main reasons is that the mechanism of human learning is still very different from that of machine learning. Both HL study and ML study are in their infancy. It’s up to us to study and improve HL and ML.

· Diffused Thinking vs. Focused Thinking

o Different from AI/ML, the human is more creative and can learn better in a relaxed mode.

o Different from traditional wisdom, relaxation also helps to learn.

o Focused and diffuse modes (relax, let your mind flow in diffuse mode, then think in focused)

o Learning something difficult takes time.

o Metaphors provide powerful techniques for learning

· Sleep on your problems

o Different from AI/ML, human needs to sleep.

o Again, different from traditional wisdom, sleeping is not a waste of time. It in facts also helps us to learn.

o It might surprise you that sleep can actually help you consolidate your learning. During your sleep, the brain tidies up your ideas and concepts by erasing unimportant memories and strengthening what you want to remember in daytime. It’s like rehearsals of what your brain did when you are awake; it traces the patterns built, and makes the connections stronger and deeper in your memories.

· Recall

o Recall can help us learn. This is kind of obvious. The key thing here is that Dr. Oakley points to Dr. Jeff Karpicke’s research about retrieval practice to provide scientific support behind taking a couple minutes to summarize or recall material you are trying to learn.

o It goes a long way to taking something from short-term memory to long-term learning.

o Even recalling material in different physical environments can help you grasp the material independent of any physical cues that your brain may have.

o Interesting enough. “recalling” method can be used to improve human learning and also AI/ML

o For instance, the picture below shows the similarity of human brain and the computer memory architecture.

o The computer can retrieve data faster if the data is used or recalled into the cache area rather than stored into the mass storage area.

o Similarly, the recommendation to prepare for a exam is to recall what you have learnt and putting the knowledge into the working memory so that you can recall them faster in exam.

· Chunking

o The idea of “chunking” is to unite bits of information together through meaning.

o Each chunk is a network of neurons in our brain, which can then be integrated with other small chunks and form bigger chunks.

o Dr. Oakley suggested a great step-by-step process to approach learning something.

o First, survey and priming — this involves scanning a book or the syllabus of a course, for example, to get a general idea of the bigger picture.

o Second, observe an example. Then, do it yourself. And, finally, do it again and again in different contexts.

o This can be a big challenge for today’s machine learning. As the computer needs to understand the context and then put the data into perspective and then form the chuck for learning and cognition. There are various research on going. Below are some examples.

o The goal of sharing here is to motivate the readers to practice chucking in our daily learning and think of new ways to improve cognitive machine learning.


  1. Coursera’s “Learning How to Learn” course