"Pitfalls & Biases In Machine Learning" With Dan Elton, PhD

Hosted By
Kevin K.

Details
Hi Everyone,
- Please join my friend, Dan Elton PhD (Physics), who is in Seattle for only 48 hours, as he gives us a talk on "Pitfalls & Biases In Machine Learning".
- • Description This talk will go over some pitfalls every machine learning practitioner should know about. Several common technical pitfalls will be discussed: overfitting, failing to clean your data (the "Schenectady Problem"), not normalizing your data, and trying to use machine learning for extrapolation. Next a few different meanings for the word "bias" will be discussed - the statistical meanings of the term, sampling bias, and social bias. Social bias occurs when machine learning models do things which perpetuate racism, sexism, inequality, and other negative phenomena in society which we do not wish to perpetuate. Following the work of Kate Crawford two harms from socially biased models can be distinguished - harms of allocation and harms of representation. Informative real world stories will be used to illustrate each type of bias. The talk will end by stepping back and taking a "big picture" view on machine learning, what it is, what it's good for, and how it contrasts with scientific understanding of the world. • Speaker Bio. Dan Elton is a postdoctoral researcher at the University of Maryland, College park, working with Prof. Peter W. Chung and Prof. Mark Fuge. He has a Ph.D. in physics from Stony Brook University. His current work focuses on applications of machine learning to molecular property prediction and molecular design.
- Dan's website is http://www.moreisdifferent.com
Dan on Twitter @moreisdifferent. - • Address: 958 111th Ave NE, Bellevue Washington, 98004. Venue: Elements Apartments, Elements Cinema.
- Come to the middle building with the escalators inside on the 1st floor. From there look for the elevator and go to 3rd Floor (Lobby, Lease Office), and to you right will be a room titled, "Cinema"
- •Parking: Parking is on P2 of the center building of Elements in Bellevue: 958 111th Ave NE, Bellevue, WA, 98004 I
- • If you message me in advance I can prepare a parking fee waiver for you. I expect to only be permitted to issue a limited number of these. • Food & Drinks: There will be casual networking after the talk. Please bring some drinks and snacks to share with the group. • What to bring:There will be casual networking after the talk. Please bring some drinks and snacks to share with the group.
• Important to know
Parking is on P2 of the center building of Elements in Bellevue: 958 111th Ave NE, Bellevue, WA, 98004
If you message me in advance I can prepare a parking fee waiver for you. I expect to only be permitted to issue a limited number of these. My mobile is 425-614-6987.
• What to bring
There will be casual networking after the talk. Please bring some drinks and snacks to share with the group.
• Important to know
Parking is on P2 of the center building of Elements in Bellevue: 958 111th Ave NE, Bellevue, WA, 98004
- If you message me in advance I can prepare a parking fee waiver for you. I expect to only be permitted to issue a limited number of these. My mobile is 425-614-6987.

Bellevue Machine Learning & Artificial Intelligence Meetup
See more events
Elements Apartments
958 111th Avenue Northeast · Bellevue, WA
"Pitfalls & Biases In Machine Learning" With Dan Elton, PhD