Joel Oskarsson

@[email protected] | PhD Student @liu_universitet | Probabilistic Machine Learning |

[Christoph Molnar 🦋 christophmolnar.bsky.social @ChristophMolnar]

Modeling data in unfavorable terms and GIFs.

Let's start.

Reinforcement learning - Build things. Robots! Agents! Game AIs! Just make sure it works in toy environments. And ONLY in toy environments!

[Joel Oskarsson @joel_oskarsson]

I am happy to be attending the Nordstat conference this week 😄 Will give a presentation about Deep Gaussian Markov Random Fields for general graphs on Wednesday.

[NORDSTAT 2023 @nordstat2023]

#nordstat2023 is starting today! We have ~300 participants, 171 talks and 40 posters.
It is hosted by @StatsChalmersGU https://t.co/iS1qUCvowY

[Naomi Saphra @nsaphra]

Regular reminder of the best mathematical resource in machine learning, The Matrix Cookbook. Don't know how anyone ever does any math without it. https://t.co/HtaDcu7DCT

[Joel Oskarsson @joel_oskarsson]

This approach additionally allows us to make predictions for arbitrary times in the future. For more information see the paper (https://t.co/DKU7gM8A4o) and my recorded presentation (https://t.co/CiUOpoBgz6).

[Joel Oskarsson @joel_oskarsson]

Next week I'll be attending AISTATS in Valencia, presenting our paper "Temporal Graph Neural Networks for Irregular Data" in the second poster session. In this paper we tackle the problem of irregular observations in graph-structured time series, considering irregularity ...

[Joel Oskarsson @joel_oskarsson]

both in the form of non-equidistant time steps and partial observations of the graph. Our solution to this challenging setting is a temporal GNN model defining time-continuous latent states in each node of the graph.

[Bernhard Schölkopf @bschoelkopf]

Another AI paradox: people are excited about LLMs, some even think that AGI is just around the corner. But some students are depressed how they can still get a PhD. Is it becoming pointless?
Some personal notes on this. (1/8)

[Joel Oskarsson @joel_oskarsson]

@malpractitioner I thought that just applied to the subsets, not example elements of the sets. But I have no idea 🤪 Just know that Venn diagrams should show all possible intersections.

[Joel Oskarsson @joel_oskarsson]

I very much enjoy the idea of pronouncing IEEE as a word rather than an abbreviation 😄

[Annie? Rauwerda? @anniierau]

i just found out that tumblr went to town on this venn diagram i made

[Aidan Gomez @aidangomez]

more ML code golf necessary

[Fredrik Lindsten @LindstenFredrik]

I'm looking for two postdocs to join us at the Division of Statistics and Machine Learning at Linköping University.

1. Geometric deep learning/2D materials discovery
2. Probabilistic machine learning/sequential generative models

Deadline for both positions: March 19

[Randall Munroe @xkcd]

Square Packing https://t.co/B5XJgwPHy3

[Nathan 🔎 @NathanpmYoung]

god is dead and the most efficient way to pack 17 squares into a square killed him

[Randall Munroe @xkcd]

Coordinate Plane Closure https://t.co/8I0PaonexX

[ProbAI = Probabilistic AI ✌️ @probabilisticai]

ProbAI 2023 applications are now open! Join us on June 12-16 in Trondheim 🇳🇴. → https://t.co/dEA7VBgZZe #ProbabilisticAI #ProbAI

[Matt Henderson @matthen2]

watch a 2 layer neural network learn to separate two classes to the left and right

[Finnish Center for AI 🦣 @[email protected] @FCAI_fi]

Today, FCAI and @WASP_Research are building new collaborations. A delegation from Sweden 🇸🇪 is visiting @CSAalto and will join AI Day tomorrow! @samikaski is opening the joint poster session
cc @ELLISforEurope

[Philipp Schmitt @philippschmitt]

New research-y project: Blueprints for Intelligence, a visual history of artificial neural networks from 1943 to 2020
https://t.co/CcPFYIxeJz