Alex Zakharov

PhD student at @UniOfOxford. RL and world models.

[Alex Goldie @AlexDGoldie]

1/ 🪩 Automating the discovery of new algorithms could unlock significant breakthroughs in ML research. But optimising agents for this research has been limited by too few tasks to learn from!

Introducing DiscoGen, a procedural generator of algorithm discovery tasks 🧵

[Hamza Adnan @_hamzaadnan_]

My first paper with @JacksonMattT and @az_prd is out on arXiv!

Introducing MaskLAM: a simple, architecture-free tweak that makes Latent Action Models (LAMs) focus on agents ONLY: 4× higher returns, 3× better latent actions, and a step towards learning from unlabelled videos! ⬇️

[Alex Zakharov @az_prd]

Come join us at Friedman 102!

[Zafeirios Fountas @zfountas]

Excited to announce our #RLDM2022 workshop "Temporal Representation in RL" on 11 June @BrownUniversity ⚡️

We have an amazing lineup of speakers discussing time in biological & artificial brains incl. @RichardSSutton, Marc Howard and @criticalneuro!

See: https://t.co/kszIwFSV7q

[Alex Zakharov @az_prd]

Come check out our new paper at ICLR2022! 🚨 We introduce a hierarchical generative model for videos, which continuously controls its hierarchical depth for processing.

[Zafeirios Fountas @zfountas]

⚡️Our new work with @az_prd titled "Variational predictive routing with nested subjective timescales" (#VPR) is presented at #ICLR2022 tomorrow 28 Apr at 10:30 am — 12:30pm BST.

https://t.co/f7fdWrbDLh

VPR is an event-based hierarchical generative model for videos… 🧵👇

[Alex Zakharov @az_prd]

@RevolutApp my account has been locked for two days now, saying “We need some information”. Nobody in the support chat is replying to me..

[Alex Zakharov @az_prd]

@mpd37 @ChengSoonOng @NeurIPSConf Wow, this looks awesome

[jared tumiel @jnearestn]

How do you learn about something exciting, new, and interdisciplinary? A syllabus of course!

Today I'm releasing my own version of a syllabus to go deeper on the Free Energy Principle and Active Inference https://t.co/AP4LF5ss36

[I.Share @IAMNERDIShare]

I didn't go to work because a cat fell on my head and then a dog with shoes tried to do me justice.

[Amir Alansary @amiralansary]

Join us this afternoon to check our @Grail2020 @MICCAI2020 work age prediction using @DevelopingHCP brain surfaces presented by my student @az_prd

Thanks @DanielRueckert @schuhschuh & all coauthors from @BioMedIAICL

Paper https://t.co/mimB1LrdBO
Code https://t.co/EdVlbxiuQU

[Alex Zakharov @az_prd]

@strangetruther @MaCroPhilosophy @mpshanahan @zfountas The system is trained via variational inference. Simply put, (1) links states to actions by attempting to mimic the predictions produced by the model-based MCTS planner

[Alex Zakharov @az_prd]

@strangetruther @MaCroPhilosophy @mpshanahan @zfountas Thank you! In short, 1) habitual network acts as a model-free component mapping states directly to actions, 2) encoder infers a hidden state corresponding to a particular observation. For more details, please see this recent work: https://t.co/bANR0PrBNN

[Zafeirios Fountas @zfountas]

2 things to share! 1) A few days ago, our paper on Deep active inference with MC methods got accepted @ #NeurIPS2020🥳 2) We released this exciting work that connects DAI with episodic memory & subjective time resulting in the following very cool features for model-based agents:

[Matt McGill @MattMcGill_]

Preprint Time!
Episodic Memory for Learning Subjective-Timescale Models https://t.co/dEzTSw5X8v

Basic idea: Store only "interesting" memories and learn dynamics between them. Plan actions to get to interesting states skipping the boring stuff.

This has many cool properties..1/6

[Sergey Levine @svlevine]

Thanks Lex for the fun discussion a few months back!

[Notion @NotionHQ]

Inline math equations have been a top feature request for us for quite some time. They're finally here!

Render beautifully formatted equations in-line with your text, using the full capability of the KaTeX library 🧮

Guide & FAQs here: https://t.co/j5q7Xl5kYu

[Lex Fridman @lexfridman]

Here's my conversation with Karl Friston, one of the greatest neuroscientists in history. From the vodka and wine in the background to the mathematics of existence, life, intelligence, and consciousness, this was a truly fascinating conversation. https://t.co/eOvpbAY8Dh

[rapha @rapha_gl]

Have you ever asked yourself "how did research get done before LateX?!"

Oh boy do I have the thread for you… (1/?)

[Matt McGill @MattMcGill_]

We're running our first (completely online quarantine-friendly) Animal-AI child study! If you have a child aged 6-10 that might enjoy solving our puzzles please fill in this form and we will be in contact https://t.co/JBGfZodz4F

There's a £5 amazon e-voucher for your time!

[Alex Zakharov @az_prd]

@FrancisRhysWar1 @lexfridman @USCLONI Wow, looks amazing