Pattern Recognition and Problem Solving
·Project prakash :
using the real world to guide the machine.
Tapan Sir said that dont try to develop the machine but try to perceive the real world and then try to interpret that with the machine learning Like he did. He tried to understand the Color and gray scale comparison and tried to map the results of C2C, C2G, G2C and G2G to real world. how the results can be used to interpret the reality
I asked Sir how can we use the multimodal learning to improve computer vision? my main idea was that Video has in itself the audio modality in built how can we use the video understanding with audio. If I try to incorporate the the idea that Prof Tapan has given. The idea is same as it was used in modality bias by Piyush Pasi from IIT Bombay where he derived a reason of modality bias when incorporating the audio with the video and solved that modality bias problem with the help of machine learning.
No chatgpt cant help me solving this issue. I have to solve it myself where its about understanding that the sometimes you dont have to find a problem but you have to observe the problem. Thats what I was trying to do with balanced multimodality.
If I try to find a problem in TAL wrt audio. I have to ask what is TAL. It is find the starting and ending location in the video of a particular event. Now I was trying to find a problem where I can find the usage of audio in TAL I felt there is very much use of audio in TAL but how. I cant find the problem by thinking more but I have to find the problem by changing my perspective to find the problem. First of all the question is do I really know or understand the problems that are already solved or how multimodality is incured in TAL. If I read a single paper a day then also its ok. It should be ok. I have to come back again to access this ability to evaluate how am I thinking and in what direction should I go next.
What is the difference between problem solving in engineering and problem solving in research?
Thats what should the change persepctive till now I was trying to do phd from the perspective of a learner. I have to change that perspective from being a learner to a problem solver like above. and solve a problem in real world I have to learn to observe. It is about learning but not like learning as a student to learn a concept tp pass the exams. But to learn to focus your onto a problem for change in the real world. You cant be an observer by always indulging in abstract perfect world. The world is imperfect and I have to accept that one fact in real life. Your life, your research can be abstract from real world problems yes delulu is solution but not all the problems are solved using delusion.
Give yourself exercise everyday to observe a concept in abstract world and connect it to real world. Give yourself a task to observe. Observe the results of your decisions. Observe the problems connecting to the abstract concepts. Like how the TAL you use in real world. Surveillance is an example what are the other appplications where TAL is useful.
Second read a paper a day to complete the TAL and other tasks. Learn to recognise the patterns. Train your brain to recognise the patterns in real world…in concepts…in problems.
Learn how to find out pattern recognition in the world. The thing you are trying to practice using excel is about pattern recognition where I am procrastinating.
This should be the goal of the second year is to train yourself for pattern recognition in the research. Then only you can go on further to work on a paper and to publish one.
What is difference between pattern recognition you understand till now and pattern recognition you are trying to learn. I recognise pattern in data. Like finding out what is distinguishing object. And thats how we are taught. The pattern recognition that I am trying to learn is not about the data. Data is infinity. It is like finding a needle in the haystack. Its not a positive example. But tuning your mind to recognize the pattern instead of depending on the data for the pattern.
Dont let today’s evaluation go waste.