Dr. T.J. Jankun-Kelly

Blog post + preprint motivating use of statistical decision theory/info economics & why you need to provide study participants sufficient information to optimize whenever goal is to evaluate human decisions or predictions. W/@AlexKale17 @jasondhartline statmodeling.stat.columbia.edu/2024/01/30/min…

I won’t be able to follow #ieevis live next week, so hopefully I’ll see some good summaries. Been missing those.

@charles_perin *Now

@charles_perin No I’m really sad I’m not in person.

Redesigned my #gfx lighting notes. Now I start with a high level rendering equation and then show how we simplify simplify to get things like Phong and PBR. Given the recent move towards lighting realism, I think this is a good direction.

@_mcnutt_ @JessicaHullman @ieeevis @nicolaskruchten See, my laziness pays off! 😆 But, really, thanks. Nothing on the page pointed to it that I could see.

New post on Multiple Views, just in time for @ieeevis! @nicolaskruchten presents an eval framework for vis notation (eg, declarative grammars, DSLs, other toolkits). The metrics complement qualitative methods like Cognitive Dimensions of Notation @nicolaskruchten/6c03b9292780" target="_blank" rel="nofollow noopener">medium.com/@nicolaskrucht…

@dalbersszafir It does, and drives up attendance. The lack of reasoning other than “expense” vaguely is disheartening.

@dr_tj Here's hoping we can reboot hybrid in the future :-) It's a ton of work, but it expands our reach and inclusiveness significantly.

I'm not sad that I'll miss #IEEEVIS (I've missed one or two before); what I'm sad about is that we had a great online option and now we don't.

I miss #BeOS for some reason. macOS feels the closest, but BeOS was fun.

Academia is mostly people telling you that you aren't good enough and sometimes there's a lunch

@EliCDavis So, not sarcasm as much as lazy. 😆

@EliCDavis Ed’s (Dr Swan) dissertation was all about splats in the context of volume rendering; lots of others built on that. I haven’t dug into the current hype, so not sure what’s what (other than compute power). web.cse.msstate.edu/~swan/publicat…

So how are these new #gaussian #splats in #gfx different from the splats of two decades ago?

I don't think the point of an adaptation is to retell it moment for moment, that just does a disservice to both art forms by pretending that they involve exactly the same skills, which obviously they do not

You applied for a #PhD / postdoc position, but the professor does not reply to you. Why? As many students are concerned about the lack of response, let me try to explain this phenomenon point by point: ▫️ 1. In industry, recruiters and HR take care of the selection process. They are hired specifically for this. BUT in academia, it is the professor’s job to find and hire the right person (and find time for that). ▫️ 2. A typical faculty is HEAVILY overloaded by duties. It’s like being a CEO of a small company + HR + recruiting + R&D lead at the same time. As a result, faculties are often struggling with finding time to reply to each candidate. ▫️ 3. Many candidates ask for a response (and feedback) and expect engagement from the faculty. Now, imagine you are a professor. You’re getting 5-15 emails per HOUR from colleagues, collaborators, funding agency, admins, students, etc. But you can dedicate only 1-2 hours per day for ALL your emails. You will barely have time to scroll through your mailbox! ▫️ 4. The number of candidates can exceed 300-400 per position, with many emails from candidates who expect replies. So, as a result, professors often reply IF the candidate is of interest to them. Of course there are those who find time to send a short message to everyone, but such people are statistically rare. ▫️ So, my advice: 1️⃣ Apply and don’t wait for a reply. Submit your application to the groups that you would like to join. If you don’t get a reply, you can send a follow-up email but don’t do it many times. 2️⃣ Tailor your cover letter to the position, make it stand out. When competition is so high, you want your application to be memorable. 3️⃣ See the absence of reply as “no position”. And don’t stress over it. Finding the right fit takes time for all of us. 4️⃣ IF you are ready to invest into traveling: - You can say that you will in their city/town, and if they are interested, you can visit their group and give a research talk (without any obligations from their side). If they agree, it may be your chance to present yourself in person. ▫️ Unfortunately, the academic system is constantly overloading faculties. This causes a lot of misunderstanding. Many faculties wish they had that time to respond to everyone. I do my best to find this time. But it doesn’t always work out smoothly. #AcademicTwitter #AcademicChatter

After years at the cutting edge of machine learning research, I finally solved AI safety 🤖✅ I am excited to announce: SafeGPT 🧷 The safest LLM is the one that refuses to respond to anything 🙊

I'm still intrigued with the idea of **visual proofs**. Just see it & understand it. This is one of the better ones. There must be some internet catalog of similar examples. Can anyone point me to it?

Sometimes game design takes a while. #gamedev #IndieGameDev

@AlexKale17 Online? I assume not but with a check.

I’m really looking forward to this!