Following the idea that “I should focus on things that I would be proud of”, my goal for today was to dive further into the study of neural networks for image segmentation, in order to prepare myself for a project that I will be tackling in January.
Narrator: “He did not dive into neural networks that day.”
To make a long story short, today I was abducted by my procrastinating habits. In the morning I noticed that there was a new chapter in the “Ward” online serial, at the same time that my RPG group started a discord discussion about rules for our next game. That itself would not have been too bad, except that after I cleared all of that, I somehow fell into the trap of re-reading old chapters of Worm. Suddenly, it was 20:00 and I did nothing this day.
To allow myself a tiny silver lining, in those last few hours I managed to scrounge myself the willpower to work through the Keras tutorials for CIFAR-10. Most tutorials I found were really crappy, so I ended up following the basic explanation from the docs and imitating my Pytorch CIFAR-10 tutorial using the Keras framework, which ended up being more educational. Both frameworks achieve the same result with similar networks, although Keras seems to train a tiny little bit faster.
Here is hoping that tomorrow I procrastinate less and spend more time grokking my student’s image segmentation code.
New year: Time to look back at what I did in 2018, and think about how I want the next year to be different.
Thinking about 2018, I can say it was a “scramble” year for me. Above everything else, my tenure examination was looming over my head. Everything that I did, at least until September, was evaluated by “does this help me pass my tenure evaluation?”. On the good side, I did get tenured this year, and this means work security and freedom to make longer termed plans. On the downside, I feel like I did a lot of things that I am really not interested in, just because of tenure, and my research passion suffered accordingly. By the end of the year, I just wanted all these side projects to get over with.
So I guess my aspirations for 2019 is to par down on the number of side projects, and try to focus on things that I really want to grow and love. Say less yes, and focus on things that I think will make me proud down the line. Of course, I still have many leftover projects from 2018. Specially because in August I was scrambling to get funding, any funding, and surprisingly many of my proposals were accepted. Too much of a good thing? I mean, I like these projects, but there are just so *many* of them — I was expecting maybe 1 in 5 of my applications to get accepted. But hey, I should be celebrating, right? :-)
With this in mind, my goal for this year is to try to focus on things that I will feel proud of having done at some point in the future. With this in mind, my very first day (evening?) of the year was spent catching up with neural network frameworks (Keras and Pytorch), so that I can better read and comment on my student’s code. My goal by the end of the week is to: 1- Run an image segmentation model (using U-net?), 2- Run some simple werewolf AI bots, 3- get up to speed with the grad-free library of EC-based neural networks, 4- hack my linux machine to play nice with the Sony Digital paper, and last but not least, 5- Try to blog all of this stuff.
Today I went to the Vietnamese embassy to get a visa for the upcoming ICPC contest. On my way there, I fell and caught myself quickly enough with my arms. However, my left arm is still hurting. This is not the first time that a relatively minor thing makes me lose my breath or hurt for an unusual amount of time… I really need to take the time to exercise, or at least stretch…
When I returned home, a filming crew was in front of the building I live in. One of the staff was smoking right in the front door. I weakly asked him if it was okay to smoke there (it wasn’t), and while the smoker seemed abashed by that, another woman in the group told me to mind my own business, which I did… I really need to work on my aversion to conflict…
A few days ago I posted a small rant here. It was really hard to put that out, and even after the post I kept questioning if it was appropriate to “complain in public”. I feel the weight of not feeling able to take things out of my chest to other people or online… I really need to work on taking the trash out from my inner house from time to time…
I think the first time I “blogged” was back in 1999, when I was 19. I had recently joined college (UNICAMP – Brazil), and found out that I could make a personal webpage. I set up a bare bones green HTML page, added a list of favorite links, and started writing short text called “rants” (the word blog did not exist back then).
At that time, I never imagined that anyone would read what I was writing, so I poured my heart out. I can’t quite remember what I wrote though. That old webpage has fallen off the web, and even the Wayback Machine only keeps caches of that page starting from 2002.
From 2004 to 2010 I started my WordPress blog, together with my life as a graduate student. Thinking about that time, there was a lot of self promotion, trying to sell my image as a “nerd” to myself, of trying to insert myself in a conversation. It looks bad when I write it like this, but it was tremendously beneficial and fun to me. I miss those blogging days.
That said, towards the later half of those years, some things started to change. As I participated in more and more communities, online and offline, my persona started to fragment. Who I was in a forum started to separate from who I was in academia, and who I was in the Blog. As I started to join Social Media, such as Twitter and Facebook, this fragmentation accelerated.
When I started working, this fragmentation reached a peak. Every time I thought about writing something, I wondered about how that would reflect in all my other personas. Not only writing, either. Who am I when I post a game to Ludum Dare? Who am I when I write on my blog? When I post on twitter? I wonder how much of this self-censure led me away from blogging. I see a video that makes me think, but is controversial. Can I post a link to it? I binge on a computer game the whole weekend, is it safe to brag about it online? Can I talk about my fears?
When I was a kid, there was a saying that I like “do not worry about what other people think about you, they are too busy worrying about what you think about them”. And yet I worry. And yet I miss being part of the conversation, even if the conversation is only between me and all the digital ghosts of the blogosphere.
I was thinking about it after watching some video talks from youtubers that I have enjoyed recently. They talked about how to separate their online persona from their private persona when talking about controversial topics. I miss talking, I miss a lot of things.
On the other hand, I have abandoned this place for so long, that I might as well be speaking into the void here (and to Antonio). This thought kinda feels me with determination.
My gaming computer broke just before Steam’s Summer sale. The timing couldn’t be more perfect. I arrived home that Summer Friday, thinking about what games I would like to add to my backlog, and the computer didn’t turn up. No beep, nothing. Dead. It was a long summer…
This computer was my first major purchase since I arrived in Tsukuba. I had received my first salary, ever, and walked into the PC Depot at Kenkyu Gakuen. I wanted something to last, so I pushed up all the stats beyond what was acceptable at the time. I could run XCOM and Starcraft 2 at the highest settings wow!
That was six years ago, so all in all the computer lasted a very decent time. I upgraded the GPU last year to be able to run XCOM2, but other than that I never had any problem with the specs. I wonder what was the problem — I hope it was only the power source, and not the motherboard or the CPU, although I probably should change those too.
This week (24 thru 28) I’ve been to Kaohsiung in Taiwan for some Rest and Recuperating. Some thoughts off the top of my head about the trip:
It was really good to unplug. I did no gaming, youtubing or redditing, and minimal twittering during these days. My brain feels rebooted and resetted.
I also had no coffee the entire week. My last and first cups were in the airplane.
I was really impressed that every train station had great wifi. It was very useful to download last minute travel info.
I used the iPass electronic money card for all the public transportation, and it had an obscure system where from time to time it would not charge us any money. I still haven’t figured out how it worked, but I was glad for it!
We used taxis two times, and both times the taxi drivers scammed us. But other than that, the trip was hassle free.
My feed got destroyed after two days of heavy walking. I was hoping to last more than that… :-(
Chinese temples are so over the top with decorations, with hundreds of dolls in hundreds of positions everywhere. I wonder
if they are built all at once, or evolve over time.
Air China’s economy class chairs were really comfortable, but the movie selection was poor. I slept my way in, read a book my way out.
In the event “buddha meets the gods”, one of the caravans visiting the temple was playing the mickey mouse club song. Did Disney finally bought their way into godhood?
There is probably much more to say about the trip, but if I try to write everything I will never post this, so there!
I have recently read the blood Blindsight, a “hard scifi” about modified humans meeting aliens species for the first time. I don’t remember exactly, but I think Jon recommended it to me. He’s just cool like that.
The book was fun, if a bit hard to get into at first. I wrote a review thingie that you can see on goodreads through this link (Don’t need an account to see the review)
My friends in the gaming group have suggested that I start preparing another Sword and Sorcery Megadungeon.
The idea is that this would be a “backpocket” game, which I could pull out and GM whenever we don’t have
enough players for whatever the main game is at the time.
Backpocket game is a game that I can run with minimal session preparation, and with a small, un-fixed cast.
A megadungeon would in theory be perfect for this: 1) There is minimal story besides the experiences the players
had in previous sessions; 2) the characters are disposable, and (in the case of the Sword and Sorcery system)
can be done very quickly; 3) we can fill in the player gaps with retainers.
In practice, the last time I tried this it didn’t go that well — it was very hard to resist the temptation of
chaining sessions together, and running roleplay focused games. Soon, players were getting attached to their
characters, and the idea of the characters being disposable, or playing without some characters being around
became less appealing.
To avoid that, this time I will try to lay heavier house rules to avoid “town game” and adventures exceeding
sessions. The main idea is to force players to flee the dungeon at the end of every session, and have a penalty
table to roll for those groups that didn’t make it in time. You returned to town, with monsters in tow, in your
last hit point, and having dropped almost all your gold pieces, but you did return to town.
Also, I am making the dungeon more mystical and “fun-house”-y this time, borrowing heavily from Etrian Odyssey
— this should encourage the players to see the game more mechanically. I hope!
Last week I went to the Maker Faire at Odaiba, Tokyo
From what I could see, In most of the stands people were presenting their hand-crafted electronics. Half of the presenters were showing some sort of robot (My favorite was the guy with the “hand extender” – a phone would film your hand, and the robotic hand at the end of a pole would mimic the movement). Then we had the raspberry Pi hacks (and its Japanese versions). After that, a little bit of everything, like the guys who made super detailed star wars costumes.
The most interesting part, however, were the children workshops. There quite a few of them. In one, a person was helping the kids build elaborate paper planes and fly them through hoops. In another, this old dude was teaching kids how to program in Scratch. There was an “experimental” one, with multiple tools like air blowers, and the kids could do whatever their wanted.
At the end of the Faire, I saw a stand selling the Arduboy, a tiny programmable pocket console. While I certainly would love having one of these, what came
to my mind was trying to make a Pico-8 powered hand held using some mix of Raspberry pi, a small led screen, a disassembled controller, and much more experience putting electronics together than I actually have.
I wonder if there is some sort of workshop that would help me with the steps necessary to make this project…
A few weeks ago, I did a pretty neat game on Ludum Dare!
The game is called “Shy Gladiator”, and can be played here. The theme this time was “Running out of Energy”. As usual, I tried to find a unique interpretation for the theme. Most of the other entrants used the idea that the player had its own energy as some sort of limit. In “Shy Gladiator” on the other hand, it is the ENEMIES who are limited by energy. The idea is that you have to hold your own until the enemy dies out and die (the excuse is that the gladiator is too shy to attack). Most people reviewing the game liked the idea.
Also, for the first time, I participated as a group instead of individually, and it was an extremely interesting experience. You might have noticed that the graphics of the game are much better than my usual fare. This is thanks to the amazing work of Felipe who also did the music and part of the sound effects. It was a bit hard to collaborate on the one-file-per-game pico-8 cartridge, but we made do with a common dropbox folder to a “code” cartridge and an “art” cartridge back and forth. The end result was certainly worth it. Felipe used the entire PICO8 sprite sheet to create monsters, fireballs, a background and animations.
The feedback so far from the Ludum Dare has been great. Original idea and animations seems to be the strong point. On the other hand, many people found the game to be “too slow” or “passive”, because the main game play is to just sit and wait until the monsters die by themselves. I guess I must agree. In the first monster, the cyclops, there is a small bit of strategy where you can
try to stay close to the monster so that it uses its strongest attack and, if you can dodge it, die faster. But for the other
two monsters there is no such strategy. If I were to add anything new to the game, it would be an option to “taunt” the monster
and make them use a stronger attack at you, to add an element of “risk vs reward” play to the game.
There are, of course, many other smaller additions that I would like to make, such as a better title screen, and some more
special effects, but I am very happy with the final result. Please play “Shy Guardian” and let me know what you think!