Why we read academic papers?
We have learned from a lot of academic papers, approximately a couple of thousands papers in a year. Moreover, at every year, we create summaries for more than 2,000 papers in computer vision and related fields (e.g. Machine Learning, Natural Language Processing, Pattern Recognition). We have witnessed the yearly number of views for our contents is over hundreds thousand. Through the experiences, I will look back why we read academic papers as follows.
1. Grasping research trends
- Without knowledge, you could create a degraded version of existing research.
- You should know minimum trend in your research, otherwise you cannot generate a latest research.
2. Confirming the on-going research position
- What is the difference? Why do you carry out? What is better?
- These comparisons and questions lead your research theme to a sophisticated philosophy.
3. To create a next trend
- To establish a frontier in your research field.
- To be better, lead a right way.
Topics 1 and 2 are frequently taken in academic fields. On the other hand, topic 3 is quite important but have not been enoughly discussed so far. Please see the following figure, this is the word clouds from titles of accepted papers in CVPR 2016.
I guess the situation in 2016 and 2021 has been changed. However, now is still in ‘the Era of Deep Learning’. It is said that a person who comprehensively grasp the research trends can (relatively) easily create a new research topic/theme. If you can understand all of the research trends, you may focus on the main research trend. Otherwise, you might dare to set up a research topic outside the mainstream. In any case, the more knowledgeable you are, the better your research will be promoted! We would like to polish our knowledge from now on as well.
