Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: What interesting problems are you working on?

news.ycombinator.com

Post reply on HN
Print view View original

251–260 of 633 posts

Re: Ask HN: What interesting problems are you working on?

#251

Post by mrdmnd »

hazrmard wrote: post #180

I am working on fusing classical control with reinforcement-learning based control in drones. The good thing about RL is that it is adaptive and proactive - it tunes control from data as it streams and it learns to act with delayed feedback in mind. The bad thing is that it can be sample inefficient. It needs lots and lots of data. That is not good in a time-critical case such as when faults occur. I've come to enjoy…

I bet the folks at Skydio would want to talk to you.

Re: Ask HN: What interesting problems are you working on?

#252

Post by mtlmtlmtlmtl »

O__________O wrote: post #217
mtlmtlmtlmtl wrote: post #196

Earlier quoted context omitted.

MuZero doesn't, but from what I know MuZero was only on par with AlphaZero, and while AlphaZero beat Stockfish initially, Stockfish has since caught up and continues to beat LeelaZero, an OSS implementation of AlphaZero in TCEC.

My understanding is that MuZero beats AlphaZero, but not by a significant amount. Even if Stockfish is better, to me at this point feels needless, given it’s highly unlikely a human over significant volume of matches would have any hope of beating any of them. It is possible that new approaches for masters might be found, but for average player, guessing it’s highly unlikely that anything new might be uncovered. Am I…

You're not missing something. I just wanna see how far we can take this thing, you know? 4000 ELO seems within reach now. What about 5000, 6000 etc?

Computer chess is essentially a sport unto itself at this point, beating humans is sort of uninteresting, though I'm sure grandmasters will always find uses for ever stronger engines. They're mainly used to study openings, and since analysing from the start of the game is extremely complex, stronger engines are always helpful.

Re: Ask HN: What interesting problems are you working on?

#254

Post by yakshaving_jgt »

My colleagues and I are writing a system for insurance companies to prepare their reinsurance submission pack data, which they would otherwise spend enormous amounts of time and money doing in Excel manually.

We have some pretty big clients using this in anger, and they’re wild about it.

It’s not glamorous work, but it’s important!

Re: Ask HN: What interesting problems are you working on?

#255

Post by techstrategist »

iLoveOncall wrote: post #141
Ilasky wrote: post #113

I’m working on making body doubling a more main stream approach to accomplishing everyday tasks. Body doubling is known within the ADHD community and entails performing a task in the presence of another. More details here: https://www.healthline.com/health/adhd/body-double-adhd It helps to engage motivation by using another person as the proxy. I wrote a bit about how I think it works here: https://doubleapp.xyz/blog…

I don't know if I have ADHD or not, and I don't think so because I'm sometimes able to focus for hours if it's really an end-of-the-world type of situation, but just having someone passively working next to me is 100% useless. My girlfriend is always working next to me, and that doesn't prevent me from getting distracted every few minutes. The only thing it helps is for stuff like cleaning the house where the problem…

that sounds like hyperfocus, which is absolutely part of ADHD.

Re: Ask HN: What interesting problems are you working on?

#256

Post by feoren »

ximm wrote: post #144

I got deep into color contrast algorithms. Human vision is a rabbit hole indeed. You would think that this is a well established field. But then I found claims in online discussions (including at the W3C) that were obviously wrong on a mathematical level. The more I looked, the more the arguments fell apart. So instead of reading online discussions I concentrated on reading the actual formulas as well as research on…

Your blog post makes lots of sense if all you care about is one threshold: contrast must be above X for readability. Then the equivalence-under-monotonic-function makes total sense: you literally don't care what the function does above or below your threshold. This extends to if you pick N threshold points (text must be above Contrast1, contextual borders between Contrast2 and Contrast3, etc.) by first looking at the function and then "eyeballing" those thresholds. But is that the best we can do? The difference between (Ymin / Ymax) and log(Ymin / Ymax) starts to matter a lot as soon as you start talking about "twice the contrast" or "20% more contrast", doesn't it?

Re: Ask HN: What interesting problems are you working on?

#257

Post by cxr »

philip1209 wrote: post #168

I'm working to reinvent personal websites. My goal is to replace social media with personal sites + newsletters, which are more distributed, privacy-friendly, and calm. https://postcard.page The product seems simple, but there are so many opportunities for optimization. There's caching, CDNs, custom domain support, dynamically-generated open graph images, image optimizations, email sending, email reputations, analyti…

I've always thought that plans where custom-domain-or-not is a differentiator between free and paid are unethical. In this case, antithetical, too. (If a sufficient number of people abandon Twitter as part of the your-own-personal-homepage bandwagon, but they make the unfortunate mistake of letting their dart land on Postcard and signing up for a free plan, then you really haven't achieved the stated goal. You've recreated the way she was tethered by name to twitter.com/alice, except the difference is that you're the beneficiary, not Twitter—because she is now held captive at alice.postcard.page.)

Previously: https://news.ycombinator.com/item?id=21921176>

Re: Ask HN: What interesting problems are you working on?

#258

Post by O__________O »

provability wrote: post #237

To reduce the adverse effects of fake evidence (eventually called deepfake), off and on since 2004, and especially since ten years ago, I've been thinking up, NOT DETECTION methods, but prospective, pro-active methods to instrument a data stream, to increase the odds that video, audio, or other time-series data (including e.g. scientific experimental data) eventually can be proven to have been created from real life…

To me, being able to trace the province of content is much more dangerous than deep fakes, since such systems have the potential to become wide spread, unavoidable, and constrain free exchange of information.

Re: Ask HN: What interesting problems are you working on?

#260

Post by russellbeattie »

Ilasky wrote: post #113

I’m working on making body doubling a more main stream approach to accomplishing everyday tasks. Body doubling is known within the ADHD community and entails performing a task in the presence of another. More details here: https://www.healthline.com/health/adhd/body-double-adhd It helps to engage motivation by using another person as the proxy. I wrote a bit about how I think it works here: https://doubleapp.xyz/blog…

This is great! I have ADHD and find this helps a lot. A YouTuber I follow has a two hour video called "Work with me" of just him working at his desk with music playing. When I first saw it, I laughed. But then one day I was having trouble staying focused, remembered the video and put it on my TV and I found it to be surprisingly effective! Unreasonably so! So good luck with your project! You're on to something.

https://youtu.be/vEqJAReXiN4

Post reply on HN