My career is based on the microsoft .net/azure stack, and that's what I am doing for my current client. A lot of it is just crud/batch import stuff for collecting/validating data on various models which includes a lot of business logic built around complex relationships in their data structures.
On the side, I am working on a sports web app with a friend as I find time using the same stack.
Also, lately I have been trying to learn reinforcement learning. What about yourself?
Automated deployment systems where solar powered mesh network repeaters with two camera channels each (SBCs) come online and handshake/pair then push auto-configurations, SSH, lighttpd for IP camera images, etc. Then go to secure mode to lock it all down and configure it for operating mode.
Edit: think in trees, weatherproofed with solar panels
Power HARVESTING and having a downtime/idle hours is key in my experience.
I like the Banana Pi Pro because it has onboard wireless and two camera channels built in. People often get third party bolt-ons that further complicate power consumption issues.
Working on my macOS image viewer app, VoidView. I hadn’t been serious about polishing it but now I’m adding some simple quality of life improvements, especially for making artists that use apps like Bridge or PureRef for displaying references happier.
Trying out a hardware side project this time. Figuring out a way to track your surf board or kite board once it's lost at sea during a session. It's a pretty interesting challenge.
I've got a number of things I would like to work on that are more into the realm of hardware. One thing I'm really wanting to work on is a way for me to manage and interact with my cars ECU through a web interface. Would be useful for a number of things, such as keeping a better maintenance schedule, useful statistics, maybe even tuning...Seems pretty fun.
As far as tracking goes, I can't think of a need for it with my idea, but I know it's used a lot commercially already.
It's a proposed classifier algorithm that is suppose to be better than Random Forest and XGBoost for classifying high dimensional data. The data sets are cancer data (prostate and myeloma). Unfortunately it's not going to be publish because I'd like to graduate sooner and that the software does not meet certain criteria for the journal we were aiming for.
The proposed algorithm uses two technique:
1. My forest consist of GUIDE decision trees by Dr. Loh. It is better than CART and M4.5 and such because it does not have the selection bias problem. CART and M4.5 are bias to selecting categorical predictor for node splitting. They're also bias on variables that enable more splitting so decision trees usually contain more levels. GUIDE is also aim at finding interactions candidate to split if it is statistically significant.
2. CERP by Dr. Moon. It makes the trees within the forest less correlated among each other. Much more so than Random Forest. Accuracy takes a hit as your correlation gets higher (obviously zero is the best). It also enabled ensemble of ensembles (ensemble of forests). Of course some of you may state that you can do ensemble of random forests but it is naive and won't help you.
I should be defending in this month or next month.
Not until my advisor okay my paper. He wants to see it first.
> Can your approach be used to do feature segmentation of 3D (triangle mesh) data?
Sorry, I have no clue what this mean so I don't think this is my area of expertise nor can I answer this properly. I'm an master student for applied statistic, this seems to be a computer science machine learning question?
I can point toward papers regarding either GUIDE or CERP if you like.
Going from zero musical ability and knowledge of music theory to composing and playing jazz, classical and hymn. I've given myself a year to do it and documenting my progress every day: https://write.as/poseur-to-composer/about
Cool, it takes time. I picked up the electric guitar, again, I am aiming to be a decent guitarist in ten years. I currently play bass in a hobby metall band which is fun.
This was my weekend project but I'm still working on it in my spare time.
The concept is simple:
All news stories have a ticking timer. After it expires the news story is gone forever. There is also a clear now button at the bottom that destroys all the stories on the screen if you want to binge even faster.
Once the story goes away, it's gone forever. You will never see it again! So it's like the HN homepage that never shows you the same story twice.
Made with Vuejs and hosted on Aws lambda (serverless)
There aren't many options because I just uploaded the first version and everything is still pretty rough. It's just a concept site I made for my own amusement :)
This is really cool! I quit my big tech job a year ago, too, so I can definitely relate to the joy of focusing on software centered around something you're passionate about.
I'm teaching myself graduate mathematical analysis using Rudin and whatever I can find online. It's a really beautiful subject, and I'm finding myself look forward to the little time I have in the day for it.
I wish to resurrect the "spirit" of the fox/dbase family where operate on data was much more natural. This lang also mix some ideas of array langs like kdb+.
If my bet is correct, this will erase the need of ORMs yet make it work as easy.
Is build on rust, a lang I also learning along the way (making it even MORE slow to progress!!!) but I start to get the gist of it.
sounds interesting. Considering posting updates when you design or evaluate concepts or ideas about the internals, designs, and philosophy behind its abstractions (I guess there is more than tables related to tables and some syntactic sugar on top)
I'm currently attempting to bootstrap taxammend.com (a service to send your 1040X IRS tax form easily) and gearing up for amiaccessible.com (a service to run automated accessibility checks against your website, enabling good-faith efforts are improving internet accessibility).
I am also currently working on my Master's in Cybersecurity and Info Assurance.
working on my crippling depression / ADHD, unemployed (i don't want to work anymore in anything) 6 months , 0 productivity. learned a lot of category theory, functional programming and c++ tho.
Sorry to hear that but good that you are working on it. Life will look better to you in the future if you keep at it. Psychotherapy worked for me in the long run (1+ years) - suffered from severe depression for more than two years.
Hang in there. You're definitely not alone, and it does get better. I'm happy to listen if you ever need someone to talk to who has been there, and in important ways is still there.
As others have said, don't be so hard on yourself: that you're teaching yourself useful stuff is the opposite of zero productivity. In fact, you're probably learning faster than if you were doing it while employed.
My best suggestion is that your local library will likely allow you to download an app such as Libby that will allow you to listen to audiobooks for free. During my downtime, nothing has brought me more respite from my depression than listening to - no joke - about 200 books. I alternate between sci-fi and non-fiction. I never would have imagined myself a listener, but I was very wrong.
are you lifting weights and exercising? If not get on it ASAP. Revert back to the physical-activity-heavy lifestyle of our ancestors and you'll feel better
It's not as simple as "X will make you lose weight or not." Caloric deficit depends on both your intake and expenditure. You can out-eat almost any exercise regime so a somewhat healthy diet and reasonable portion sizes are always going to be important.
As someone who recently discovered unaddressed ADHD has been holding me back so much for so long, as well as experiencing that numbness of depression too, you are not alone and I have felt at least some of what you are experiencing. I wish you all the best in working through it.
Start a journal to track your diet and any pertinent symptoms. After a week or two, when you have a baseline picture of your current diet, look up some articles on research connecting diet and depression.
Off the top of my head: Certain oils are pertinent -- some should be avoided, others sought out -- and iron deficiency can cause depression.
I hope things get better for you. The hardest part is taking that first step. I've been there and I know how hard it is to even get out of bed. It's cheesy, but what really helped me was listening to motivational tapes like Les Brown/Jim Rohn/Eric Thomas. I'd just go for a walk while listening to them and over time it got a little bit easier.
I have been there too. When you are injured or ill, your duty is to survive. You can flourish and achieve when you are healthy again so don't go on holding yourself to what you can expect in other contexts.
I am working with the people at cleverpet and the small hackerpet community on making this pet training device acessible for DIY additions and new ways to train your dog/cat
I'm working on Grid engine 9, which should release some big improvements to out-of-the-box multiplayer support. It's the only source code-based Lua game engine that I know of with first-class multiplayer support that has a dogfood project. Everything else out there is roll-your-own w/ LuaSockets. I've been writing it for the last few years due to not being able to find a solution that fit my needs.
I used to use Lua all the time. It's the language I learned after QBASIC and what I spent most of my teenage years writing. I'm definitely going to stay up to date on Grid. This looks neat!
I feel most of the newsletters that aggregate stories and articles have too links. The reader gets overwhelm with what to read and what to skip.
I am taking a different strategy and aggregating only four links per week. The reader could quickly scan the four links. If any of them is interesting, click and read. Otherwise, hit the delete and move on for the day. I don't want to reader to be captive if he finds the links do no add value for them. Their time is to valuable and I respect that.
Also I am using this side project as motivation to consume the various articles for my day job.
I'm a solo developer and find it gratifying to set aside a half hour at the end of each week to write down what I did. Otherwise all the work just becomes a blur. I wrote a longer blog post about the motivation behind it here: https://mtlynch.io/status-updates-to-nobody/
I have had a pretty incredible urge lately to work on some similar software. I'd love an app, for example, like Committed (linked below) that had SIMPLE live-collaboration functionality. Figured I would demo that pretty soon, make a site, and see if it could gain some traction. Looking forward to seeing how this progresses.
I have a "lab book", but I stopped writing into it. It mostly contains information i gathered about a particular section of the code base or how a solution might work or what I learned about our business domain.
But over time the constant context switching of topics I had to work on made me stop it.
I would like to revive it. However, i want the information in there (the lessons learned this week) to be less project/issue specific, but more transcending and transferable to new problems/environments.
Do you have a set of questions you pose yourself whose answers guide the write-up and make it a worthwhile knowledge vault?
Oh, that's interesting. I do some introspection when I'm writing my weekly updates[1], but not to the depth that you're describing. But What Got Done should work as a lab book, although I haven't implemented search yet, and it sounds like that might be important to you.
The time when I do deeper reflection is when I write my monthly retrospectives. [2] I don't have a standard set of questions that I ask myself, but that's a good idea. My method so far has been to state what lessons I learned, evaluate how I did against my goals for the month, and define goals for the subsequent month.
559 comments
[ 2.1 ms ] story [ 323 ms ] threadOn the side, I am working on a sports web app with a friend as I find time using the same stack.
Also, lately I have been trying to learn reinforcement learning. What about yourself?
Edit: think in trees, weatherproofed with solar panels
Power HARVESTING and having a downtime/idle hours is key in my experience.
I like the Banana Pi Pro because it has onboard wireless and two camera channels built in. People often get third party bolt-ons that further complicate power consumption issues.
https://github.com/ESWAT/voidview
As far as tracking goes, I can't think of a need for it with my idea, but I know it's used a lot commercially already.
It's a proposed classifier algorithm that is suppose to be better than Random Forest and XGBoost for classifying high dimensional data. The data sets are cancer data (prostate and myeloma). Unfortunately it's not going to be publish because I'd like to graduate sooner and that the software does not meet certain criteria for the journal we were aiming for.
The proposed algorithm uses two technique:
1. My forest consist of GUIDE decision trees by Dr. Loh. It is better than CART and M4.5 and such because it does not have the selection bias problem. CART and M4.5 are bias to selecting categorical predictor for node splitting. They're also bias on variables that enable more splitting so decision trees usually contain more levels. GUIDE is also aim at finding interactions candidate to split if it is statistically significant.
2. CERP by Dr. Moon. It makes the trees within the forest less correlated among each other. Much more so than Random Forest. Accuracy takes a hit as your correlation gets higher (obviously zero is the best). It also enabled ensemble of ensembles (ensemble of forests). Of course some of you may state that you can do ensemble of random forests but it is naive and won't help you.
I should be defending in this month or next month.
Not until my advisor okay my paper. He wants to see it first.
> Can your approach be used to do feature segmentation of 3D (triangle mesh) data?
Sorry, I have no clue what this mean so I don't think this is my area of expertise nor can I answer this properly. I'm an master student for applied statistic, this seems to be a computer science machine learning question?
I can point toward papers regarding either GUIDE or CERP if you like.
GUIDE definitive paper: http://pages.stat.wisc.edu/~loh/treeprogs/guide/aoas260.pdf
All other papers on GUIDE: http://pages.stat.wisc.edu/~loh/guide.html
CT CERP: http://www.ams.sunysb.edu/~hahn/psfile/aiim_moon.pdf
This was my weekend project but I'm still working on it in my spare time.
The concept is simple: All news stories have a ticking timer. After it expires the news story is gone forever. There is also a clear now button at the bottom that destroys all the stories on the screen if you want to binge even faster.
Once the story goes away, it's gone forever. You will never see it again! So it's like the HN homepage that never shows you the same story twice.
Made with Vuejs and hosted on Aws lambda (serverless)
There aren't many options because I just uploaded the first version and everything is still pretty rough. It's just a concept site I made for my own amusement :)
https://splits.io
http://tablam.org
I wish to resurrect the "spirit" of the fox/dbase family where operate on data was much more natural. This lang also mix some ideas of array langs like kdb+.
If my bet is correct, this will erase the need of ORMs yet make it work as easy.
Is build on rust, a lang I also learning along the way (making it even MORE slow to progress!!!) but I start to get the gist of it.
I am also currently working on my Master's in Cybersecurity and Info Assurance.
https://m.youtube.com/watch?v=KcLz3yAqqyI
And my volumetric lightning effect too
https://m.youtube.com/watch?v=DdUve22wlvg
(Both on the unity asset store as “physics based flames” and “physics based lightning”)
As others have said, don't be so hard on yourself: that you're teaching yourself useful stuff is the opposite of zero productivity. In fact, you're probably learning faster than if you were doing it while employed.
My best suggestion is that your local library will likely allow you to download an app such as Libby that will allow you to listen to audiobooks for free. During my downtime, nothing has brought me more respite from my depression than listening to - no joke - about 200 books. I alternate between sci-fi and non-fiction. I never would have imagined myself a listener, but I was very wrong.
https://www.coursera.org/learn/happiness
Just put your routine in a FOR-loop and grind for increased STRength and STAmina stats lol
Active VR games are a legitimate cardio workout. Sites like https://www.vrfitnessinsider.com/reviews/ have real data if you don't believe me.
As for duration, I've seen people (including myself) happily play Beat Saber for hours who don't last 10 miserable minutes on a treadmill.
Off the top of my head: Certain oils are pertinent -- some should be avoided, others sought out -- and iron deficiency can cause depression.
http://hackerpet.com http://clever.pet
I'm working on Grid engine 9, which should release some big improvements to out-of-the-box multiplayer support. It's the only source code-based Lua game engine that I know of with first-class multiplayer support that has a dogfood project. Everything else out there is roll-your-own w/ LuaSockets. I've been writing it for the last few years due to not being able to find a solution that fit my needs.
I feel most of the newsletters that aggregate stories and articles have too links. The reader gets overwhelm with what to read and what to skip.
I am taking a different strategy and aggregating only four links per week. The reader could quickly scan the four links. If any of them is interesting, click and read. Otherwise, hit the delete and move on for the day. I don't want to reader to be captive if he finds the links do no add value for them. Their time is to valuable and I respect that.
Also I am using this side project as motivation to consume the various articles for my day job.
https://whatgotdone.com/
I'm a solo developer and find it gratifying to set aside a half hour at the end of each week to write down what I did. Otherwise all the work just becomes a blur. I wrote a longer blog post about the motivation behind it here: https://mtlynch.io/status-updates-to-nobody/
(forgot to add the link!)
https://chrome.google.com/webstore/detail/committed-newtab-t...
>like Committed (linked below)
I can't find the link you're referring to. Did you mean to add a link to the bottom of your post?
I would like to revive it. However, i want the information in there (the lessons learned this week) to be less project/issue specific, but more transcending and transferable to new problems/environments.
Do you have a set of questions you pose yourself whose answers guide the write-up and make it a worthwhile knowledge vault?
===
Not sure if I expressed myself in a coherent way.
The time when I do deeper reflection is when I write my monthly retrospectives. [2] I don't have a standard set of questions that I ask myself, but that's a good idea. My method so far has been to state what lessons I learned, evaluate how I did against my goals for the month, and define goals for the subsequent month.
[1] http://whatgotdone.com/michael
[2] https://mtlynch.io/retrospectives/