How do we learn? Nobody knows for sure. We might know how we learn to play basketball (it takes a lot of practice) or how we learn to write computer programs (it takes a lot of thinking); but we don’t have a definite answer to an effective learning strategy for generic things including college education. Learning and teaching are the two sides of the same coin – you can not judge teaching effectiveness without taking learning effectiveness into account.
The fire-making analogy I mentioned in today’s class. Here is the complete version (I know my oral communication is not good enough so here is another try). Imagine you are Fred Flintstone and you want to learn how to make sure. Teacher A spends one month to teach you all kinds of sophisticated skills about making fire by drilling on dry woods: he shows how to tell the dryness of a piece of wood, how to spin the drill, how to observe the process and so on. Teacher A is a great instructor with patience and skills. Now there is another Teacher B – he does not teach you the practice of drilling on woods; instead he tells you to think all possible ways of making a fire – how to do it with a piece of ice instead; how to exploit the resource in nature (e.g., lightening strikes); how to search for flint stones which are a kind of stones you never heard before except in your lastname. After one month, you have not mastered the skills of drilling on woods yet; but do you think you have learned less from Teacher B? This Teacher B might even tell you to forget about the skills of making fire because what is more important than making fire is sustaining fire. My goal is to become that kind of teacher.
It is really a different perspective toward learning. Some people call it “research-based learning”. I like this term but I think here “research” is not just research in the narrow sense (what PhD is about) but a more general way of curiosity-based problem solving process. What is research-based learning? Let me define research first. It is RE-search. How do we search for things we need? A century ago, we went to library; now we Google it on the web. Personally I have always felt the success of Google is not just some smart business decision or good marketing strategy; it is because search is a fundamental tool needed by humans. The capability of search for knowledge is as important as the capability of harness energy for any civilization in the universe. If you are still in the passive mode (to be fed the knowledge you need), you are already lagging behind your peers (a good learner actively searches for what fits their need).
How do we search? I am afraid this is like the question “how do we learn?” No one has figured out the machinery of neocortex yet so I can only tell you the best practice from my personal and empirical experience. The most important tip (in my opinion) seems to be organize what you have known. You have learned calculus; but why do we need it? Who invented it? Newton – you said. But why did Newton need to invent calculus for his mechanics? If you don’t know the answer, do you know how to search on wiki? You could start from reading Newton’s biography or calculus’ history. You will know calculus is a tool invented to study changes in the physical world. Then wait – suppose you were studying the problem of edge detection and you know edge is also intuitively defined by some kind of changes. So can calculus be used for detecting edges? Bingo! If you were thinking of this question before 1980, you would make your name attached to the various edge detection operators we learned in this class. That is precisely the process of how knowledge is created. You all can do it by searching, thinking, searching, thinking … and finally making connections. I think this procedure is at the foundation of both scientific research and active learning.
Hello, Prof. Li. I really like your blog. It gives me a lot of inspirations.
Now I am pursuing the PhD degree. I have a general question about the PhD research. What kind of qualities do you think a good researcher should possess? For myself, I was a good learner, but I find it is more difficult to become a good researcher. Everyday, I spend a lot of time reading papers. However, it seems I am just following others’ thoughts. I often ask myself why I cannot have these ideas. Besides, there is a feeling in me that I am in a brimless ocean and don’t know where is the land.
Thanks very much.
BTW, my research direction is computional photography.
Hi, Zhimin – Thanks for visiting my weblog. Your comments reminded me the old days of my PhD study. Just like you, I also read tons of papers in the library (there was no Google) but found it mostly a waste of time. Looking back, I think the following tips might be helpful to your case: 1) try to develop your own vision about your field first. Taking computational photography as an example. Do you know its history? As far as I know, it is still a relatively young field. How did this field manage to catch increasing attention in recent years? Where is this field evolving? How is this field connected to other scientific and technological fields? Many of these questions – you can solicit opinions from authorities in this field, but more importantly, you must do your own reasoning to understand the evolutionary law of your field. 2) understand where your talent is: theory or practice. For computational photography, I would guess practice weights more because your work will be sold to graphics or vision community. Therefore, hand-on experience is critical in this business from what I have seen. During PhD training, I think the most important step is to obtain some initial success, which will help you build up your confidence. With one little step forward (e.g., a conf. paper), you will be able to build on top of it or explore new territories.
Hi, Prof. Li. Thanks very much for your advices. They are quite helpful. Maybe I should start to rearrange my thoughts.
Another question is how should we read others’ paper/work. What should be of most concern?
Thanks again.
Have you thought about: why do we read others’ papers? It is possible to do research without reading papers at all. One famous example is John Nash – when he was a graduate student, he used his own mental power to reinvent many things from the scratch.But most of us do not have that kind of mental power, so we need to stand on giants’ shoulders. But whose shoulders? A century ago, such question is easy to answer because good science was still easy to identify (journal publication did not even have peer-review system we have these days). Rapid advance of technology had unexpected negative impact on science – every year, so many papers were published and it has become more and more difficult to sift out pearls from sands. Indeed, top-ranking CS conferences such as SIGGRAPH already have such sifting process. So you are lucky to work in the field of computational photography (it has not been contaminated – I hope so), but meantime making progress in such emerging field requires vision (see my previous reply). My suggestion is: pay more attention to high-level questions than technical details. For example, how did this paper get accepted by SIGGRAPH? what new killer application has it proposed or advanced? what are the remaining open issues or questions? It is usually more difficult to propose a new problem to work on than to polish an existing solution. At the early stage of your career, it is a good idea to meet the easier goal first (this goes back to the confidence issue I addressed before).
Thanks very much. Your suggestions are quite helpful to me.
To be honest, I am jealous of your students. I am no longer a student. Remember my school time, all my teachers are not A-type teachers, nor B-type teachers. Although slow, but I can learn something, which I want to know, from book or web.
BTW: the Research-Based Learning, in my mind, like another ancient famous wisdom in china from Confucius (China’s greatest teacher). He said that “ju3yi1fan3san1”, and I really do not know how to express in English.