My Purdue Experience - Dimitrios

 I am now officially done with the internship and I can look back at the whole experience with a different set of pairs, as the hard times are over. I will start by saying some things that I am proud of and that I and the rest of the team have achieved over the course. When I first started working with Deepak,   he asked me to help him perform an ultrasound quality assessment, a task I had never even remotely seen or heard again in my life. I am proud of the fact that I could adapt to this highly demanding environment and I could commit myself to a project I did not choose. I am proud that I could keep up and read the literature, which is partly due to the experience GSTR 210 and 310 had equipped me with. Without these two, I would not be able to distinguish which sources are important and prestigious. At the same time, in GSTR 310 I learned how to use Zotero, which was a decisive tool for the whole internship experience. I am also proud of the coding performance I showed, b...

 This is my final entry post :(  I wanted to start this one by saying how much I appreciate the TAs. Because they genuinely try to give you feedback or help you in the comments, this makes it so much different than just doing a task and forgetting about it. Having someone to read what you went through is fantastic. I could try and explain what I did today to my mom, but the thing is that most of my tasks are so technical, and she would never understand them. She would still be very happy about me but learning to express and explain your work, in other words, sell your work to an audience that can understand you make all the difference.  Today I finally got done sorting the code, and I collected everything in the repository, with a code fully functional and logically sorted. Deepak said that he would take over the draft corrections of the paper and that I could rest assured that he would submit everything on Sunday. The only comment we have gotten so far about the research...

 I woke up sick today, so I asked Deepak to take the day off from work and have a conversation about the experience and what I should take away from it or what I should try to improve. We had a very long conversation, but I will try to summarize some good and bad experiences we talked about. During my internship here, so many smart people were introduced to me. I made good friends with other interns while I was doing my internship. Most of the time, I felt stupid, but being in a room with people who are better than you is a fantastic experience. I acquired cultural sensitivity as people practiced and held to a wide variety of beliefs. It was both tough and interesting. As a student, I was always carefree about my responsibilities and assignments, just like any other student, but when you enter a job environment, things move quickly. One assignment after another is assigned to you, and clearly, you cannot put them off. You are gradually schooled to function and advance in that setti...

 Today I worked on standardizing and fixing my code. Inside the paper we are submitting, we have included a link where everyone interested can find our code, organized and fully functional in all operating systems. Now, these two words sound easy, but I left the office today with a headache, as it is tough for me to locate the correct files and code to add them to the repository. The reason I am struggling so much is that when I was writing it, I left minimal comments, and I have so many files with the same name but with a number in the end, like cluster.py and cluster1.py. At the time, when I was experimenting with the code, it made sense, and I thought I would not forget it in 3 weeks when I would write the paper. Comments and good names are seriously the best things a programmer can do to their selves, which is something I learned in both 224 and 236, but I never used it in any of the two courses. Back then, I was not penalized, but now I am since it's taking so much more time t...

 My framework figure was accepted!!! Now, I have already submitted the first complete draft to Deepak and the professor as of today. The Conference deadline for the paper submission is the 30th, and my internship ends on the 29th, which is perfect timing. Since I have submitted the whole paper draft, I can now say that I am proud of the writing in that paper; although counting all the smaller subdrafts, I probably have written more than 25 pages of information. After Deepak's comments, I cut these pages to 8 with a very heavy heart. I always had a problem with feedback because I always think that my work is amazing and there is no reason to delete or rephrase a significant part of it. This is a behavior I noticed I had after my GSTR 310 professor pointed out that my second draft was my first draft but with a different name. Since then, I have tried to accept feedback more open-hearted. I think this time was a bit different since when I am writing an academic paper in Berea, I usual...

  Minimizing Disbelief is the single most important principle I have learned from my research experience. I did learn a lot of useful skills, but skills can be taught, how to sell your work is an art and can be grasped only with the proper influence. The professor told us that when Ph.D. students at Carnegie Mellon were asked what is the most important skill they acquired during their PhD, they said the art to sell your work, but I believe that selling your work comes from a core principle that every researcher should have, minimizing disbelief. It’s not only researchers, in fact when writers are writing movies, like Star wars, if you actually sit down and think about the core idea, but it’s also kind of silly if not abstract and possibly dumb I would say. But why, I and so many millions more are fans of star wars and enjoy the film? Because it doesn’t matter if we start with the plot from the beginning to the end, it’s the storyteller who has in ways minimized our disbelief that t...

 For the past 2 days, I have continuously been working on the network architecture; I never realized this would be such a tricky part because I already knew the structure, and it seemed very simple and straightforward how it works. I wasted about 5-6 hours on it last week, I thought I had a final version. The architecture is essentially a framework and is kind of describing my whole work in one figure. After discussing this with the professor about my paper, it became evident that there were some major flaws with the design I already had created in adobe photoshop since the loops were not visible and the professor could not distinguish the different parts of it. Here is the thing that really bugs me with writing a research paper. I can write a bunch of crap with even spelling mistakes and it will make sense to me, until someone is right there with me, which is when I realize that I have not been careful enough. This is kind of making me anxious about the final version, because wh...

 I am not yet done with the paper writing, but I have done a significant process. I never knew photoshop would be so helpful in presenting plots and creating charts, and I am grateful I have learned so much in this internship. As I said before, I appreciate being outside of my comfort zone, but some things, I get asked to do jobs that are impossible for me. If there are two things that I have used in this internship and was not taught enough in my courses in Berea, this is Linux and ROS. Linux is everywhere and is the most common OS for computers in the research field; I do not understand why we have not been introduced enough to it. We did get to use MobaX for an assignment in CS236, and since I started my internship here, there wasn't a day I did not interact via it with Linux. As for ROS, I had never even heard of it before, but apparently, it's the most widely used OS for Robotics and, therefore, a valuable skill to add to your resume, especially pursuing a CS career. Now, ...

 Two days later and I am already four drafts in. I've always preferred quality over quantity in my strategy. I selected a subject that piqued my attention and located a qualified mentor for the same. I then started to work slowly but steadily to lay the groundwork. I eventually learned the truth about a problem and began to consider with the necessary technicality and pragmatism. When I was able to identify the correct answer, I adequately recorded it. To grasp what has been done previously and the open and significant difficulties in that subject, you must first study and comprehend the earlier work in your field. Typically, the first two years of a Ph.D. program are where you begin to understand your thesis question. When I first got into the internship, for the first weeks, I was led to believe that I had to find a way to perform quality assessments on medical imagery because they needed it for the completion of the Ph.D. Some weeks later, after I presented my work to Deepak, he...

 I was very busy this week, and I still am, so I forgot to write my entry. I am kind of happy I have to do the entries because I always wanted to start writing a journal, and I always had no motivation. I am sure I will be reading some of these entries years into being a Ph.D. student or my career and laugh at my little problems. These problems for now, though, are very big for me. I need to submit the paper until Sunday, as the conference is taking place at the end of this month, and this is when they will be publishing it. I decided on the network architecture but never realized how many results I needed until I started writing the paper. I will be adding the very first and roughest draft of all on this blog post so that you can see how many details I need to add. I started with the methodology, which was a suggestion by Deepak. Today I finished writing the methodology and the math behind the logic of the US processing but realized that I have gathered not only just a few results...

 Sometimes I think I am so close to the tail, but other times I think my work is just garbage and easy to emulate for a Ph.D. student. For the whole day yesterday, I thought that the final architecture was going to be an Autoencoder with a loss based on comparing the original image (we call that the ground truth) and the reconstructed image, depending on the loss would change the weights. Now, I have realized that this loss would be great in guiding the Autoencoder in extracting features that represent the image but not actually clustering them depending on their quality. After realizing that, I had a depression attack for most of the day. I started looking for ways to implement the clustering loss on the encoder loss to guide the network towards understanding quality. Briefly, quality images one, two, and three look very similar to the human eye. Still, they are very different in quality because of specific features you can see in the image, like a liver or kidney boundary. At the...

 Today, Deepak told me that there is a conference where we can submit our work on ultrasound quality assessments, but we need to finish the research by Friday. That means I need to sum up all the work on unsupervised learning and submit it to the network, as well as miscellaneous contributions on the topic. As I mentioned in my previous entry, the experience I had at Cornell really helped me think about my research from a different point of view and gave me some creative ideas about improvements. Now, I would love to try all of them out, but doing research is a multidimensional challenge. Time is the most important resource we have in our hands, and between winning and managing it could make a difference between a Nobel prize or not. I am not doing Nobel prize-worthy work, but I need to take into account that I cannot try all the things I wish I could. Based on that, I have decided to keep the preprocessing and postprocessing of the data I get the same and focus on trying different...

 I am still not back at the research lab at Purdue. Rather I work online, so I cannot tell you many technical details or conversations I had in the lab. Still, I can talk a little about some thoughts I have been dealing with the last few days, reflecting on my work up until now and on how the research experience has given me a new view on the Computer Science field and my major. After all, the importance of internships comes from reflecting on them and learning what you like, what you dont, and getting curious about other things.  I am not sure if what I am working on right now is called engineering, computer science, or information science. I feel like it is informational science, and if so, I am pretty sure this is what I want to do later on. The tours I got at Cornell, as well as the professors and department heads I have spoken to, have been saying the same thing, that the field of computer science is very interdisciplinary. My understanding is that the goal is for student...