Makis Ntouskos (@mdouskos) profile picture

Associate Professor @UniMercatorum specializing in #AI, #ComputerVision, #DeepLearning and #Robotics; Co-founder and research director @WorldUAVHellas

Joined July 2012
Photo by Makis Ntouskos (@mdouskos)Photo by Makis Ntouskos (@mdouskos)Photo by Makis Ntouskos (@mdouskos)Photo by Makis Ntouskos (@mdouskos)Photo by Makis Ntouskos (@mdouskos)Photo by Makis Ntouskos (@mdouskos)Photo by Makis Ntouskos (@mdouskos)Photo by Makis Ntouskos (@mdouskos)
Makis Ntouskos retweeted
Exciting news! MegaSAM code is out🔥 & the updated Shape of Motion results with MegaSAM are really impressive! A year ago I didn't think we could make any progress on these videos: shape-of-motion.github.io/re… Huge congrats to everyone involved and the community 🎉
16
160
16
1,073
95,600
Makis Ntouskos retweeted
This paper seems very interesting: say you train an LLM to play chess using only transcripts of games of players up to 1000 elo. Is it possible that the model plays better than 1000 elo? (i.e. "transcends" the training data performance?). It seems you get something from nothing, and some information theory arguments that this should be impossible were discussed in conversations I had in the past. But this paper shows this can happen: training on 1000 elo game transcripts and getting an LLM that plays at 1500! Further the authors connect to a clean theoretical framework for why: it's ensembling weak learners, where you get "something from nothing" by averaging the independent mistakes of multiple models. The paper argued that you need enough data diversity and careful temperature sampling for the transcendence to occur. I had been thinking along the same lines but didn't think of using chess as a clean measurable way to scientifically measure this. Fantastic work that I'll read I'll more depth.
123
284
78
2,226
388,812
Makis Ntouskos retweeted
Check out the @ramones_eu project as they test their prototype instruments in the hydrothermal field of Milos Island! 🌊🔬🧑‍🔬 ramones-project.eu/ramones-t… #oceanexploration #marinebiology #hydrothermalvents #research
4
7
621
“Choosing a distribution is not a statistical question; it is a modeling decision. No statistical test can tell you whether a particular distribution is a good model for your data” - yes!
In the last week, three people on reddit/r/statistics have asked about testing whether a sample came from a Gaussian distribution. The answer is that you should never test for normality. The result is a non-answer to the wrong question. allendowney.com/blog/2023/01…
3
24
3
211
43,975
NEANIAS organized the Atmospheric-workshop at the Department of Physics of the National and Kapodistrian University of Athens, with the aim of presenting the work that is active in the field of open science and in particular the atmosphere. Keep reading at https://nitter.gq/t.co/t74RwrJn4C
Makis Ntouskos retweeted
CVPR: We welcome new students! Students: Cool! I just submitted my very first paper! CVPR: You now also need to review papers. Students: Not sure how to do that. Here is my attempt. CVPR: Your reviews are too unprofessional. You are now banned for two years. Students: ...
Voting is now closed for the motions proposed at the #CVPR2022 PAMI-TC meeting. All motions have passed: Motion 1: Withdrawn Submissions Will No Longer Be Made Inaccessible to Reviewers Yes: 867 No: 354 Motion 2: Making Authors Responsible for Reviewing Yes: 677 No: 553
14
44
9
369
#CVPR2022 Motion 3: "Any reviewer who has accepted an invitation to review but violates the reviewing guidelines set forth by the conference will be prohibited from submitting any papers to CVPR for up to two years." My vote is a resounding NO! 1/n
9
25
14
273
Makis Ntouskos retweeted
So many transformer explanations want me to intuitively understand what's happening. I just want to understand what the code does. This is 25 lines of pure functional code which helped me, and might others. github.com/awf/functional-tr…
6
87
3
564
Finally somebody run a nice ablation on all the (Ne)RF-class training acceleration papers! (InstantNGP, DVFo, Plenoxels) This is a *must* read
ReLU Fields: The Little Non-linearity That Could @AnimeshKarnewar, Tobias Ritschel, Oliver Wang, Niloy J. Mitra tl;dr: grid-based 3D representation + ReLU are almost NERFs, but much faster. arxiv.org/abs/2205.10824.pdf
3
15
2
130
The workshop will start in a few minutes. NEANIAS Underwater team will discuss and demonstrate novel, cutting-edge cloud services tackling critical aspects of underwater mapping tasks. #OpenScience Join the Workshop at https://nitter.gq/t.co/FYkW7VJYTM
1
7
1
12
Over the last half century, clinicians and researchers have made enormous advances in detecting and treating melanoma. The common element is clinical data: images, histopathology results and information from clinicians and patients. https://nitter.gq/t.co/kj36k1uHr1
9
1
10
Today is #EarthDay2022 !! NEANIAS aims to offer innovative digital solutions for the underwater world, the atmosphere and space that help us to know and care for the planet. Let's have a nice and sustainable world!!
The goal of the full-body scanner is to take hundreds or thousands of high-resolution pictures covering the whole body of the patient in one go and create a full avatar in which the doctors can see the details. Learn more about the iToBoS optics challenges https://nitter.gq/t.co/ihsY6FviPd
We are happy to announce the 2022 edition of ICVSS! Applications are officially open. ICVSS 2022 - Call for Participation Computer Vision: a renaissance? Sicily 10-16 July 2022 Application deadline: 19 April 2022 ICVSS Web: iplab.dmi.unict.it/icvss
1
14
2
52
New Year 2022 is coming!
2
2
f2f or virtually... but working together for a happy and successful 2022 !!!
2
1
16
1ος ΠΑΝΕΛΛΗΝΙΟΣ ΔΙΑΓΩΝΙΣΜΟΣ ΦΩΤΟΓΡΑΦΙΑΣ ΜΕ ΧΡΗΣΗ DRONE:Η World UAV Federation σε συνεργασία με το περιοδικό PHOTONET και τη φωτογραφική κοινότητα PHOTO CONTEST διοργανώνει τον "Drone Photo Contest 2021" Ενημερωθείτε για τις λεπτομέρειες του διαγωνισμού:nexusmedia.gr/drone-photo-co…
1
4
Makis Ntouskos retweeted
… and here is a single list of all #CVPR2021 papers cvpr2021.thecvf.com/node/142 Thanks @pvkhoi for pointing this out!
4
19
ECCV is finally A* ... that previous A rank was a problem before (also: ICLR, see below).
Wow, @iclr_conf receives CORE Rank A* 🥳 It is very important for PhD defenses and academic labs in many universities! Happy to contribute a little bit 😊 @hugo_larochelle @srush_nlp portal.core.edu.au/conf-rank…
4
5
2
42