Keegan Hines

422 posts

Keegan Hines banner
Keegan Hines

Keegan Hines

@keeghin

AI&cyber research @microsoft, adjunct professor @Georgetown, co-founder @CamlisOrg. alum @wlunews, @UTAustin, @sfiscience. monoids, fixies, yes and.

Katılım Nisan 2014
802 Takip Edilen453 Takipçiler
Joshua Saxe
Joshua Saxe@joshua_saxe·
One topic for discussion: the feeling that if the public discussion of AI is this confused how can I trust any other public discussion of any complexity
English
4
0
8
328
Joshua Saxe
Joshua Saxe@joshua_saxe·
Needed: a support group for people working in AI trying to make sense of how the arcana of their work is suddenly the stuff of cocktail parties and cable news banter
English
3
0
11
931
Keegan Hines
Keegan Hines@keeghin·
Registration is open for CAMLIS 2022! Two days of awesome talks on Cyber+ML and featuring two keynotes this year from Amanda Rousseau and Mikel Rodriguez. Join us in Arlington or join the live stream for free. Register now, limits spots for in-person! camlis.org
English
0
9
19
0
Keegan Hines retweetledi
Gabriel Peyré
Gabriel Peyré@gabrielpeyre·
Oldies but goldies: N Metropolis, A Rosenbluth, M Rosenbluth, A Teller, E Teller, Equation of State Calculations by Fast Computing Machines, 1953. The (wrongly attributed …) Metropolis-Hasting algorithm samples from a density known only up to a constant. en.wikipedia.org/wiki/Equation_…
English
3
36
196
0
Bayan Bruss
Bayan Bruss@cbbruss·
My daughter decided to join for one of our summer research fellow’s final presentations and got to hear from the totally brilliant and bad*** @jinxmirror13
Bayan Bruss tweet media
English
3
1
6
0
Keegan Hines
Keegan Hines@keeghin·
ARTIFICIAL NEURAL NETWORKS ACTUALLY ARE ONLY SLIGHTLY SIMILAR TO BIOLOGICAL NEURONS. THERE'S AN AMAZING AMOUNT OF DIVERSITY AND COMPLEXITY IN THE BIOLOGY (DENDRITE STRUCTURE, NON-LINEAR CONDUCTANCES, HOMEOSTASIS) THAT THE DEEP LEARNING PEOPLE AREN'T EVEN PUTTING INTO THEIR MODELS
Keegan Hines tweet media
English
1
0
31
0
Keegan Hines
Keegan Hines@keeghin·
This week at AIES2022, Arthur Research Fellow @ziebrah is presenting her paper "Equalizing Credit Opportunity in Algorithms". If you work in ML and financial services, check this one out! arthur.ai/blog/arthur-re…
English
0
0
5
0
Keegan Hines
Keegan Hines@keeghin·
he's a 10 but he's a frequentist...
English
0
0
1
0
Keegan Hines
Keegan Hines@keeghin·
Working on anomaly detection but worried about "contamination" anomalies in the training data set? Cool idea from arxiv.org/abs/2202.08088 - treat as a latent var problem and estimate whether each item is anomalous or normal, while also training AD model
Keegan Hines tweet media
English
0
0
2
0
Keegan Hines
Keegan Hines@keeghin·
Breaking: In a 6-3 decision, the Supreme Court has declared that correlation does imply causation.
English
0
0
3
0
Keegan Hines
Keegan Hines@keeghin·
We're thrilled that @itsArthurAI will be presenting three talks at ODSC this week. I'll be presenting about high dimensional data drift, @jessicadai_ will be talking about Fair ML, and @karthik6d will be talking about counterfactual explanations. Hope to see you there!
Arthur@itsArthurAI

This week, three members of the Arthur team will be speaking at @ODSC East in Boston, MA about drift detection, operationalizing fair ML, and more. Hope to see you there! Click here for more info: hubs.la/Q018gcVS0

English
0
0
5
0
Keegan Hines
Keegan Hines@keeghin·
Loving this amazing talk from @BasicScienceSav about geometric deep learning in particle physics at LHC. 🤯🤯🤯Thanks for hanging with us today at @itsArthurAI !
English
0
1
4
0
Keegan Hines
Keegan Hines@keeghin·
what is utilitarianism on trasnfinite sets?
English
2
0
1
0
Keegan Hines
Keegan Hines@keeghin·
Counterfactual explanations are an exciting development in explainable AI. In our recent AAAI paper, we present an efficient and flexible technique for computing counterfactuals in high-scale settings. @karthik6d @Sahil1V @johnpdickerson
Arthur@itsArthurAI

In our newest blog post, learn how to build and scale novel reinforcement learning algorithms using state-of-the-art technologies. You'll also find some starter code to get you going: hubs.la/Q016nzmC0

English
0
7
10
0
Keegan Hines
Keegan Hines@keeghin·
@rharang anomalous isn't malicious and malicious isn't anomalous, so what do ya do
English
1
0
2
0

Keşfet

@joshua_saxe @krvarshney @karthik6d @Sahil1V @cbbruss @jinxmirror13 @ziebrah @itsArthurAI