Photo by Tim Mossholder

A listing of my articles in deep learning reinforcement learning & computing

Welcome to my articles on Deep Learning, Reinforcement Learning, and computing in general. The purpose of this article is to give a road map of the articles I wrote so far. This will create a more organized index. The first part contains the major series I wrote and the second part contains a listing of articles. At the end of this article, I will answer some questions regarding the basic guideline, credits, and policy for this blog.

Series

Article Listing

Linear Algebra

Github

Credits & Guidelines

AI is coming. In some perspective, it sounds like “winter is coming” in the Game of Thrones. A lot of people talk about it but we barely realize its meaning yet (at least not until the last season of GOT). Making technologies simple and easy is the first step to proliferate them. We need to connect people to the technologies beyond meaningless buzz words in the TV ads. Many people have reached out to me for the ideas they have. Majority of them are amazing. I wish I have the bandwidth to follow up the ideas in more details.

To your surprise, my blog was not intended to discuss AI. It is a playground for me to think through some Deep learning or AI problems I have. Along the way, I may explain somethings that may trouble someone. I am glad that some people find them helpful. While I rarely (close to never) promote my articles, the readership increases so significantly that some extra cares may be warranted. Balancing between substance and reader engagement is not easy. I personally feel providing many external links throughout an article is very distractive and counterproductive in showcasing the research. I try not to explain technology as a historian. When everything pops, nothing pops. I suggest readers paying special attention to the reference section for the source of information and the original research papers. The links in the external figures act as another mechanism to trace the source of information. This approach is less thorough than typical research papers. But I find that strikes a better balance for the broader target audiences and I will review this again as needed. But if you are a researcher in the specific topic, I strongly refer you back to the original research paper for a more thorough reference.

Lady Gaga said social media is like the toilet of the internet. So allows me to do some plumbing works for a second. In average, I ban about 3 people per year from my blog. This is probably about 0.0005% of the readers. When Lady Gaga said social media is like the toilet, she probably thinks about a small group of people that getting un-proportional attention.

Let me quote the CEO of Goldman Sachs, David Solomon, directly.

The other thing I’d point to that’s so important is there is a real emphasis when people are interviewing around academics and I.Q. I think it’s way overweighted,” Solomon told the students. “There should be equal emphasis on E.Q. and how you interact with people, how you relate to people, and how you connect with people.
And even in a world where more and more machines and technology are disrupting the way we do business and the way we all connect, a lot of the value is still created through human interaction and the ability to motivate people or get people to move in a direction with you. All of those are things in my experience here that had a meaningful impact on me and helped me along the way.

I resonate a lot with this comment. From the hiring to the development process in the high-tech, we emphasize a lot on intellectual characters. However, many arguments we put forward are strongly influenced by our own experiences. Without proper exploration and validation, we just get stuck in opinionated discussions. In product design, the skill of empathy is the key to user-focus products. In some perspectives, we have an issue of empathy rather than the lack of intellectual capability.

I always appreciate people pointing out errors in the articles. Some (or many) of them are written in a very short time. Without a peer review, the errors in an article are higher than an academic paper. But sometimes, without knowing your background first, it is hard to recognize whether it is a misunderstanding or a mistake. If you can state the error and the suggested changes, it will help me to figure it out better. I often have problems when people try to write the equations in plain text. My suggestion is to save the original equation image in the article and reposit it in the comment. Or write the equation in a paper and take a picture. You can just drag and drop an image into your comment. However, a very small minority (a single digit of readers) may become very judgemental, authoritarian or having temperament issue. Don’t treat mistakes like the end of the “technology” world. Good engineers and researchers should verify information from different perspectives. If the articles can lower the barrier of technology, I am happy. But please play due diligence when you need the information for a more serious purpose.

Problems can turn ugly when you are not showing your face in the virtual world. Fortunately, this is an extreme minority for now. But it does take up time to handle them. However, I do want to discourage behaviors that may not be appropriate. My intention is to bring up some awareness here rather than being a supercop here. So enjoy your readings.

I do my best in responding to comments but less with private notes. One of the common mistakes is asking questions too broad or too vague. Some questions sound like a full project request. In reality, I take this type of questions very seriously but I find it too hard to detail the response appropriately. Also, it will be very helpful if you review the question first before sending it out. In average, I need further clarifications in about 1/3 of the questions. Last, troubleshooting is never a show-and-tell business. If you are struggling, it means you are learning. There is no short cut. Don’t expect to have your questions answered if you don’t dig around. Be comfortable to play around any open source code. By avoiding digging a hole for myself, I will be very selective in answering those questions.

As a side note, I always pulse the interests of the articles by the number of claps or the response. It helps me to determine whether further reviews are needed. Sometimes, people contact me in using or translating the materials in the blog. I really appreciate that because it helps me to understand what is important. This triggers me to review the proper credits to other resources again. If it is a translation, I often ask people to put a link for your translation back in the comment section here. In general, I am as generous as many researchers or academia. I need you to find the cures for many diseases!

Be creative and be responsible. Hopefully, you are the person that brings the next major breakthrough.