Looking at this November is going to pass again, but my website has no idea of updating at all. Although there are not many things to do in November, there are not many that can be written out, so this one will take the recent situation to make a rough count.

Recent developments

Let's start with the latest. Just yesterday (November 23, 2019), I went to take part in the test of class B operation certificate of amateur radio in Beijing, and I would have been ready for more than half a year. Class B operation certificate can be tested six months later, so the radio license I got last August can be tested as class B in February this year, but I don't know why, the association has been dragging the class B test to yesterday. In fact, the exam itself is not difficult. I listened to the lecture before the exam, basically even if I reviewed it again, and I saw the question bank itself intermittently for half a year. Last week, I spent a week doing every question, and the problem was not big. As for the method of recording the question bank, I chose to sort out the article "notes on the preparation of class B amateur radio operation certificate (v1171031 As I am a undergraduate majoring in computer science, physics and circuits are not bad, so I just choose the key points. If the foundation is poor, I suggest to take a look at the question bank (for example, I am not familiar with Ohm's law, DC power calculation and capacitance inductance volt ampere characteristics). In the past, I still thought the exam was very interesting. I listened to the lecture for two hours and answered the paper for ten minutes. Now it's time for the association to announce the results.

Further on, there are experiments in three courses: data structure, artificial intelligence and natural language processing. In fact, I'm more willing to use code to solve problems than to do papers. I know what can be expressed in code must be written in words. That's all. Among them, I think the most interesting is the assignment of natural language processing. As a public elective course, it leaves out the difficulty of the special elective course - duplicate thesis. This paper is "Research on the integration of Chinese word segmentation and Tagging Based on the ternary statistical model", which can be searched in Google's academic field, so there is no link. The general meaning of this paper is to mark the part of speech at the same time of word segmentation by using the labeled clinker library and ternary grammar model. It's not difficult to reproduce the paper. There are not many requirements for the mathematical basis. After all, it's based on the statistical model. The probability formula has been completely calculated, and the number after counting is taken into account. I like to use Java, so it's also used in this reappearance. As a result, it can be used anyway, but it's relatively slow. The accuracy of word segmentation is more than 80%, but the accuracy of part of speech tagging is 25%, which means it burns high fragrance. When I look at other people's papers, the lowest accuracy rate is over 98%. I think I'm still close. Let's talk about that.

After that, the furthest thing I can remember is to talk with my tutor. I forgot what I talked about, but only one task is the most clear. The tutor asked me to send two English papers before the end of my junior year. The suggested direction is machine learning. Just as the radio test is over, I plan to go through Zhou Zhihua's watermelon book tomorrow, and buy this book For several years, the cover of the book was still wrapped with the geography paper of senior one. Unfortunately, I didn't understand it at that time, but now it should be OK. However, I don't think I can publish a paper after reading a book. After all, I don't know what to write now, let alone a paper. What can I argue???

Future

Since it's said above that Zhou Zhihua's machine should be studied once, the following content may be related to this aspect. My interests have developed so widely that I decided to cut them down. First of all, the art aspect may have to be played. Although it's a great thing to play an instrument and I would like to learn it very much, to be honest, this skill has no economic benefit for me (that is, what direct benefit it has to me objectively). And I think the instrument can be learned as long as it can move. After mastering the foundation, it's enough to practice repeatedly. It's up to people whether they can learn anything from this practice (so-called opening up).

After that, I think Japanese and radio can't be put down. I want to learn more foreign languages, but less action. I wanted to learn Russian and Icelandic before, but these two are relatively difficult to learn. I am not a grammar major, and I am unlikely to spend too much time on it. In contrast, I may be more willing to spend time learning a new programming language. But I think Japanese is very interesting. It's not difficult to learn it. There are also people around who can speak Japanese. How much can I ask. For radio, B has been tested. Although I can't speak at the beginning, shortwave RTTY, psk31 and CW are my saviors. Although I'm unlikely to remember CW, psk31 and RTTY can be tested. In addition, I'm very interested in QRP (i.e. 5-watt global), and the cut in point may be here.

In addition, I may learn Haskell as a hobby, but it may not be very practical for me. It's really a hobby in my hobby, so I may not study hard. At present, the only thing that matters is to put machine learning in the door, because I don't like python, and the only thing I'm good at is Java (C/C + + can only be written, but it's not good at writing at all). Think about the compilation principle and operating system of next year, so it will be difficult to translate theory into implementation. I'll try to get started as soon as possible.

So this month's update is over. I hope to write a quality update next month. Next, everyone.