Chris Dann
@chrodan
Followers
464
Following
241
Media
0
Statuses
97
Research Scientist at Google, working on RL theory, former ML Ph.D. student at CMU with @AIforHI
Joined February 2012
So excited and honored to receive an ERC Starting Grant for the project BrainAlign!! BrainAlign will bring LLMs closer to human understanding by directly aligning them with the human brain. Stay tuned for our findings, and multiple postdoc and PhD openings in the coming years!
40
28
309
🍻 7 open positions (6 PhD, 1 postdoc) 🍻 Come work with me & my colleagues on human-interpretable neural models @neuroexplicit Unit will grow to 25 by 2025. Positions in Saarbrücken, 4km from France & excellent eclairs 🤌 Apply by May 31. Details: https://t.co/wQu4T05Epu
4
49
183
Seems currently more difficult than usual to recruit emergency reviewers. Still looking for one for ICML in the general area of policy optimization. Anyone with that expertise willing to help?
0
3
10
🚨6 open positions🚨 (4 postdoc, 1 PhD, 1 coordinator) for a new NeuroAI DFG-funded research unit Join our multidisciplinary team to uncover how knowledge is organized at different levels of abstraction in both 🧠 & 🖥️ https://t.co/Kl5Y6Li4B3 Apply by Jan 11, spread the word!🙏
neuroai-arena.github.io
ARENA: Abstract Representations in Neural Architectures
3
59
171
So proud of Mariya for becoming faculty at @mpi_sws_! I am very grateful for my first research experience 11 years ago at MPI Inf literally next door and I am excited to see all the researchers she will nurture!
1st day as faculty! Extremely grateful for my PhD mentors' Leila Wehbe, @tommmitchell & @TarrLab guidance and for my partner @chrodan's support in moving our lives across the ocean for this truly dream opportunity at @mpi_sws_ Stay tuned for our work at the intersection of 🖥️&🧠!
0
1
8
What a day! Thank you to all who joined the #DL4brain tutorial at #CogSci2022! Huge thank you to @CNeuromod for providing the multi-modal dataset for the Hands-on sessions. Shout out to student organizers @subbareddy300 and @jashnarora1 who did all the work :)
0
3
17
Looking forward to seeing you in-person and virtually at the #CogSci2022 tutorial on Deep Learning for Brain Encoding and Decoding! July 27, 8:30-4:30 ET Full schedule below.
How do we build on recent progress in deep learning for text, audio & vision to learn more about the brain? Join for our tutorial on Deep Learning for Brain Encoding and Decoding at #CogSci2022 Featuring hands-on sessions with multimodal naturalistic fMRI data from @CNeuromod!!
0
6
18
Happy to finally post my new model selection work with @chrodan and Claudio Gentile https://t.co/8PKViXlEQ4. In this work we ask whether it is possible to achieve best of both worlds and model selection rates simultaneously. (1/4)
arxiv.org
We study the problem of model selection in bandit scenarios in the presence of nested policy classes, with the goal of obtaining simultaneous adversarial and stochastic ("best of both worlds")...
1
2
13
Happy March 1st to all my fellow Balkan people. We gotta stick together 💟
1
1
32
Why are huge parts of 🧠 predicted well by the same DNN representation? Do these parts really process similar things? We propose a causal framework to infer the effect of complex stimuli on multiple brain zones. https://t.co/DiVIUTFwqa To appear @ Causal Learning & Reasoning 🧵
4
13
67
My group at MPI-SWS is looking for postdocs, PhD students, and research interns interested in neuroscience + ML/NLP. Starting dates are flexible & full funding and benefits are provided. Shoot me an email if interested! Please RT to spread the word! 🙏
19
161
316
So proud of Mariya for the strong defense and excited about her next chapters!
Defended my PhD on Bridging Language in Machines with Language in the Brain! I'm incredibly thankful for my advisors Leila Wehbe and @tommmitchell's support & the rest of my committee Mike Tarr, @tallinzen, @redpony Celebrated with a bike ride and a fitting collaborative effort🍾
0
0
9
Asking the important questions here
GatherTown question: what button do I click to engage in Street Fighter mode for especially heated poster interactions #NeurIPS2020
0
0
2
New study in @PNASNews on MOOC persistence- 2.5 years, 3 institutions (@harvard, @mit, @stanford), 250 courses, over *250,000* participants. New insights on scale, global achievement gaps, open science, & personalization. A Thread 1/ https://t.co/8kzOP7Q90H (@whynotyet, @emyeom)
pnas.org
Online education is rapidly expanding in response to rising demand for higher and continuing education, but many online students struggle to achiev...
15
135
323
Curious about behavior nudges to improve MOOC course persistence? @chrdann & I aided work lead by @bjfr @whynotyet @emyeom out in PNAS. Key findings: significant variation, context matters & MAB personalization insufficient w/std features. Inspires ?s for ML generalization!
New study in @PNASNews on MOOC persistence- 2.5 years, 3 institutions (@harvard, @mit, @stanford), 250 courses, over *250,000* participants. New insights on scale, global achievement gaps, open science, & personalization. A Thread 1/ https://t.co/8kzOP7Q90H (@whynotyet, @emyeom)
0
7
25
How many heads does multi-head attention need? Work from CMU shows that a large number of heads can be pruned at test time - in some cases even a single head is enough. New blog post by @pmichelX, edited by @mtoneva1: https://t.co/gODHYarmz3 paper:
blog.ml.cmu.edu
Since their inception in this 2017 paper by Vaswani et al., transformer models have become a staple of NLP research. They are used in machine translation, language modeling, and in general in most...
0
13
35
Risk sensitive decision policies are important in areas from robotics to healthcare. Come see @RamtinKeramati's work on quickly learning risk sensitive policies using distributional RL at the posters today #AAAI2020. https://t.co/9AS3pKW71j Jt wk w/@chrodan, A.Tamkin
0
3
53
Accountable, efficient reinforcement learning is a key challenge. My student Chris Dann shares steps towards this in a post on Policy Certificates & Minimax-Optimal PAC Bounds for Episodic RL. ICML 19 work jt w/Lihong Li and Wei Wei
medium.com
Designing reinforcement learning methods which find a good policy with as few samples as possible is a key goal of both empirical and…
1
14
99