Machine-learning-nederland November 2021

machine-learning-nederland@list.uva.nl

2 participants
4 discussions

Reminder: Thematic seminar: Samory Kpotufe, Fri November 26, 16h00
by Tim van Erven 25 Nov '21

25 Nov '21

Dear all, The talk by Samory Kpotufe is tomorrow: Our seminar speaker this Friday in the thematic seminar is Samory Kpotufe. Further below, there is also a list of upcoming talks that are scheduled for the second semester. *Samory Kpotufe* (Department of Statistics, Columbia University, http://www.columbia.edu/~skk2175/) Samory works on the intersection between statistics and machine learning, with an interest in adaptive methods, and was one of the chairs for last year's COLT learning theory conference. *Friday November 26*, 16.00-17.00 Online on Zoom: https://uva-live.zoom.us/j/85155421740 Meeting ID: 851 5542 1740 Please also join for online drinks after the talk. *Some Recent Insights on Transfer and Multitask Learning* A common situation in Machine Learning is one where training data is not fully representative of a target population due to bias in the sampling mechanism or due to prohibitive target sampling costs. In such situations, we aim to ’transfer’ relevant information from the training data (a.k.a. source data) to the target application. How much information is in the source data about the target application? Would some amount of target data improve transfer? These are all practical questions that depend crucially on 'how far' the source domain is from the target. However, how to properly measure 'distance' between source and target domains remains largely unclear. In this talk we will argue that much of the traditional notions of 'distance' (e.g. KL-divergence, extensions of TV such as D_A discrepancy, density-ratios, Wasserstein distance) can yield an over-pessimistic picture of transferability. Instead, we show that some asymmetric notions of 'relatedness' between source and target (which we simply term 'transfer-exponents') capture a continuum from easy to hard transfer. Transfer-exponents uncover a rich set of situations where transfer is possible even at fast rates; they encode relative benefits of source and target samples, and have interesting implications for related problems such as 'multi-task or multi-source learning'. In particular, in the case of transfer from multiple sources, we will discuss (if time permits) a strange phenomena: no procedure can guarantee a rate better than that of having a single data source, even in seemingly mild situations where multiple sources are informative about the target. The talk is based on earlier work with Guillaume Martinet, and ongoing work with Steve Hanneke. Seminar organizers: Tim van Erven Botond Szabo https://mschauer.github.io/StructuresSeminar/ *Upcoming talks:* Mar. 11, 2022, *Tomer Koren <https://mschauer.github.io/StructuresSeminar/#Koren>, *Tel Aviv University Mar. 25, 2022, *Nicolò Cesa-Bianchi <https://mschauer.github.io/StructuresSeminar/#CesaBianchi>**, *Università degli Studi di Milano Apr. 8, 2022, *Julia Olkhovskaya <https://sites.google.com/view/julia-olkhovskaya/home>*, Vrije Universiteit Apr. 22, 2022, *Tor Lattimore <https://mschauer.github.io/StructuresSeminar/#Lattimore>*, DeepMind -- Tim van Erven<tim(a)timvanerven.nl> www.timvanerven.nl

1 0

Reminder: Thematic seminar: Samory Kpotufe, Fri November 26, 16h00
by Tim van Erven 22 Nov '21

22 Nov '21

Dear all, Our seminar speaker this Friday in the thematic seminar is Samory Kpotufe. Further below, there is also a list of upcoming talks that are scheduled for the second semester. *Samory Kpotufe* (Department of Statistics, Columbia University, http://www.columbia.edu/~skk2175/) Samory works on the intersection between statistics and machine learning, with an interest in adaptive methods, and was one of the chairs for last year's COLT learning theory conference. *Friday November 26*, 16.00-17.00 Online on Zoom: https://uva-live.zoom.us/j/85155421740 Meeting ID: 851 5542 1740 Please also join for online drinks after the talk. *Some Recent Insights on Transfer and Multitask Learning* A common situation in Machine Learning is one where training data is not fully representative of a target population due to bias in the sampling mechanism or due to prohibitive target sampling costs. In such situations, we aim to ’transfer’ relevant information from the training data (a.k.a. source data) to the target application. How much information is in the source data about the target application? Would some amount of target data improve transfer? These are all practical questions that depend crucially on 'how far' the source domain is from the target. However, how to properly measure 'distance' between source and target domains remains largely unclear. In this talk we will argue that much of the traditional notions of 'distance' (e.g. KL-divergence, extensions of TV such as D_A discrepancy, density-ratios, Wasserstein distance) can yield an over-pessimistic picture of transferability. Instead, we show that some asymmetric notions of 'relatedness' between source and target (which we simply term 'transfer-exponents') capture a continuum from easy to hard transfer. Transfer-exponents uncover a rich set of situations where transfer is possible even at fast rates; they encode relative benefits of source and target samples, and have interesting implications for related problems such as 'multi-task or multi-source learning'. In particular, in the case of transfer from multiple sources, we will discuss (if time permits) a strange phenomena: no procedure can guarantee a rate better than that of having a single data source, even in seemingly mild situations where multiple sources are informative about the target. The talk is based on earlier work with Guillaume Martinet, and ongoing work with Steve Hanneke. Seminar organizers: Tim van Erven Botond Szabo https://mschauer.github.io/StructuresSeminar/ *Upcoming talks:* Mar. 11, 2022, *Tomer Koren <https://mschauer.github.io/StructuresSeminar/#Koren>, *Tel Aviv University Mar. 25, 2022, *Nicolò Cesa-Bianchi <https://mschauer.github.io/StructuresSeminar/#CesaBianchi>**, *Università degli Studi di Milano Apr. 22, 2022, *Tor Lattimore <https://mschauer.github.io/StructuresSeminar/#Lattimore>*, DeepMind [date to be confirmed], *Julia Olkhovskaya <https://sites.google.com/view/julia-olkhovskaya/home>*, Vrije Universiteit -- Tim van Erven<tim(a)timvanerven.nl> www.timvanerven.nl

1 0

Thematic seminar: Samory Kpotufe, Fri November 26, 16h00
by Tim van Erven 15 Nov '21

15 Nov '21

Dear all, With things moving online again, the good news is that we have an interesting speaker lined up in the thematic seminar on Friday next week: *Samory Kpotufe* (Department of Statistics, Columbia University, http://www.columbia.edu/~skk2175/) Samory works on the intersection between statistics and machine learning, with an interest in adaptive methods, and was one of the chairs for last year's COLT learning theory conference. *Friday November 26*, 16.00-17.00 Online on Zoom: https://uva-live.zoom.us/j/85155421740 Meeting ID: 851 5542 1740 Please also join for online drinks after the talk. Title and abstract: TBA Seminar organizers: Tim van Erven Botond Szabo https://mschauer.github.io/StructuresSeminar/ -- Tim van Erven<tim(a)timvanerven.nl> www.timvanerven.nl

1 0

Best Paper Award (CHF 500) of the Special Issue "Women in Robotics" [Robotics] (ISSN 2218-6581, citescore 3.5)
by Charlene Dong 04 Nov '21

04 Nov '21

Dear colleagues, *A "Women in Robotics Award" will be granted to the best paper published in the Special Issue "Women in Robotics" https://www.mdpi.com/journal/robotics/special_issues/Women_Robotics Deadline: 25 December 2021 Each award nominee will be assessed on her paper's originality, quality, and contribution to the field by the Evaluation Committee. The winner will receive a certificate, an award of 500 CHF, and an opportunity to publish her next submission in Robotics free of charge.* If you are interested in this project, please feel free to contact us via email charlene.dong(a)mdpi.com or robotics(a)mdpi.com within two weeks. Look forward to hearing from you. Best regards, Ms. Charlene Dong Managing Editor Email: charlene.dong(a)mdpi.com Robotics (http://www.mdpi.com/journal/robotics/) LinkedIn: https://www.linkedin.com/company/74926083/admin/ Twitter: @RoboticsMDPI The 2020 CiteScores of Robotics is released and it increased to 3.5 (Ranking Q2 in Control and Optimization category). Update: 2021–2022 Robotics Travel Awards (800 CHF) Application Deadline: 30 June 2022 https://www.mdpi.com/journal/robotics/awards Disclaimer: MDPI recognizes the importance of data privacy and protection. We treat personal data in line with the General Data Protection Regulation (GDPR) and with what the community expects of us. The information contained in this message is confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this message in error, please notify me and delete this message from your system. You may not copy this message in its entirety or in part, or disclose its contents to anyone.

1 0

2024

2023

2022

2021

2020

Machine-learning-nederland November 2021