Adam Trischler

Microsoft Researcher fostering literate machines. He/his.

[Adam Trischler @APTrizzle]

We’re hiring!

[Fernando Diaz @841io]

MSR Montréal is pleased to share two postings for senior researchers interested in joining our lab focused on natural language processing, deep learning, reinforcement learning, and fairness, accountability, transparency, and ethics (FATE).

[Adam Trischler @APTrizzle]

My wonderful sister is an academic too! Her field seems to have funner tweets, though.

[Meaning, Memory, and Movement @SpaceConference]

‘If you’ve tried to read Dante’s Commedia, most of you probably only made it out of Hell...’

Ancient spaces? Medieval ones? May as well do afterlife spaces too! Elisabeth Trischler is another demonstration of the scope of expertise on offer today. Truly _divine_ work. #M3Spaces

[Adam Trischler @APTrizzle]

@mark_riedl Congrats, Mark!!!!

[Adam Trischler @APTrizzle]

@ytay017 Congratulations, Yi!

[Fernando Diaz @841io]

Sue has influenced so much of how we approach information access and search. Oh right she also co-invented latent semantic indexing. nbd.

[Adam Trischler @APTrizzle]

@gideonmann This is like when Jerry convinced Elaine that War & Peace was originally going to be called "War: what is it good for?" isn't it?

[Adam Trischler @APTrizzle]

@zacharylipton Houses are assets but the belief they can only appreciate (or at least maintain) can be dangerous too. Predatory lenders pre-2008 convinced buyers they could take on huge, back-loaded mortgages because if they ever couldn't pay they could sell... Nope.

[Adam Trischler @APTrizzle]

"'Move fast and break things' is an abomination if your goal is to create a healthy society. Taking short-cuts may be financially profitable in the short-term, but the cost to society is too great to be justified."

[Adam Trischler @APTrizzle]

"Let’s stop designing the technologies envisioned in dystopian novels. We need to heed the warnings of artists, not race head-on into their nightmares."

A really important read for anyone in tech.

[danah boyd @zephoria]

Last night, I was honored by the @EFF. As I reflected on what got me to this place, I realized I needed to reckon with how I have benefited from men whose actions have helped uphold a patriarchal system that has hurt so many. Here is my acceptance speech: https://t.co/trjjC96CFL

[Adam Trischler @APTrizzle]

"...misconduct and purposeful suppression of the voices of women and so many others whose bodies and experiences exclude them from the powerful elite... as an industry obsessed with scale, these are the toxic logics that the tech industry has amplified and normalized."

[Hal Daumé III @haldaume3]

In NYC area and interested in reinforcement learning?

Come to RL Day at @MSFTResearch on Oct 3!

Free to register!

Awesome speaker lineup: Joelle Pineau, @smdvln, Christopher Amato, Mengdi Wang, Philip Thomas, @real_asli, Geoff Gordon and @FinaleDoshi!

https://t.co/cxqUDQahqd

[MSFT Accessibility @MSFTEnable]

"It's honestly quite a liberating concept because not only am I getting information about what's around me, but I'm actually getting the spatial information on where it is around me.” – Jason Fayre, @CNIB via @CTVNews
https://t.co/q37eTfj0Bn

[Adam Trischler @APTrizzle]

Disgusting, nonsensical, and totally divergent from what Canada is.

[Jonathan Goldsbie @goldsbie]

Seriously, this shit is wild. Never in my adult life have I seen a piece in a mainstream Canadian newspaper so explicitly decry tolerance and preach homogeneity and exclusion — using those exact words! https://t.co/qZ3VudNkr0

[Asia Biega @asia_biega]

Congrats to the Montreal AI Symposium @SymposiumAi organizers for pulling together such a diverse program. Montreal with its rich research scene is a great to be. #MAIS2019

[Katja Hofmann @katjahofmann]

Generalization in Reinforcement Learning with Selective Noise Injection and Information Bottleneck
\w @MaxiIgl, Kamil Ciosek, @liyzhen2, Sebastian Tschiatschek, Cheng Zhang & @smdvln

[Adam Trischler @APTrizzle]

Beautiful statement.

[Tsendsuren @TsendeeMTS]

Metalearned Neural Memory (https://t.co/R4Svhxuo7J) got accepted at #NeurIPS2019!

Here we have an agent with memory exploring the grid maze (left) and its top five memory recalls (right).

[Adam Trischler @APTrizzle]

Both papers were led by the awesome @ericxyuan. Only one was accepted at #EMNLP2019 (7/6)

[Adam Trischler @APTrizzle]

Two of our most recent papers just went online. Both investigate learning language interactively: we place agents in text-based environments (POMDPs) that require them to seek information through action. We want to move beyond "perceiving" with words to "doing" with words. (1/6)

[Adam Trischler @APTrizzle]

We hope these tasks contribute in scaling models to web-level QA scenarios, where word and phrase matching as in most MRC tasks are just the final step. Searching, sifting, aggregating, and reasoning over information, contextually and sequentially, are necessary prologue. (6/6)