Machine Learning Coffee Seminar: "Metabolite identification through machine learning"

Lecturer : 
Juho Rousu
Event type: 
Event
Event time: 
2017-01-23 09:15 to 10:00
Place: 
Seminar room T5, CS building, Konemiehentie 2, Otaniemi
Description: 

Helsinki region machine learning researchers will start our week by an exciting machine learning talk and discussion over coffee. The talks will start 9:15, with coffee served from 9:00. This week's talk is:

Metabolite identification through machine learning
 
Juho Rousu
Associate Professor, Aalto University
 
Abstract:
Identification of small molecules from biological samples remains a major bottleneck in understanding the inner working of biological cells and their environment. Machine learning on data from large public databases of tandem mass spectrometric data has transformed this field in recent years, witnessing an increase of identification rates by 150%. In this presentation, I will outline the key machine learning methods behind this development: kernel-based learning of molecular fingerprints, multiple kernel learning, structured prediction as well as some recent advances.
 
The next talk is:
30.1. at 9:15 in Kumpula Exactum D123: Jukka Corander "Likelihood-free inference and predictions for computational epidemiology”

 

Machine Learning Coffee Seminar: "Probabilistic programming: Bayesian modeling made easy"

Lecturer : 
Arto Klami
Event type: 
Event
Event time: 
2017-01-16 09:15 to 10:00
Place: 
Exactum D123, Kumpula
Description: 

Helsinki region machine learning researchers will start our week by an exciting machine learning talk and discussion over coffee. The talks will start 9:15, with coffee served from 9:00. This week's talk is:

Probabilistic programming: Bayesian modeling made easy
 
Arto Klami
Academy Research Fellow, University of Helsinki
 
Abstract:
Probabilistic models are principled tools for understanding data, but difficulty of inference limits the complexity of models we can actually use. Often we need to develop specific inference algorithms for new models (which might take months), and need to restrict ourselves to tractable model families that might not match our beliefs about the data. Probabilistic programming promises to fix this, by separating the model description from the inference: With probabilistic programming languages we can specify complex models using a high-level programming language, letting a black-box inference engine take care of the tricky details. This talk covers the basic idea of probabilistic programming and discusses how well its promises hold now and in the future.
 
The next talks are:
23.1. at 9:15 in Otaniemi CS building T5: Juho Rousu "Metabolite identification through machine learning"
30.1. at 9:15 in Kumpula Exactum D123: Jukka Corander "Likelihood-free inference and predictions for computational epidemiology”

 

Machine Learning Coffee Seminar: "Unsupervised machine learning for matrix decompositions"

Lecturer : 
Erkki Oja
Event type: 
Event
Event time: 
2017-01-09 09:15 to 10:00
Place: 
Seminar room T6, CS building, Konemiehentie 2, Otaniemi
Description: 
Starting January 9, Helsinki region machine learning researchers will start our week by an exciting machine learning talk and discussion over coffee. The talks will start 9:15, with coffee served from 9:00. The first talk is:
 
Unsupervised Machine Learning for Matrix Decomposition
 
Erkki Oja
Professor Emeritus, Aalto University
 
Abstract:
Unsupervised learning is a classical approach in pattern recognition and data analysis. Its importance is growing today, due to the increasing data volumes and the difficulty of obtaining statistically sufficient amounts of labelled training data. Typical analysis techniques using unsupervised learning are principal component analysis, independent component analysis, and cluster analysis. They can all be presented as decompositions of the data matrix containing the unlabeled  samples. Starting from the classical results, the author reviews some advances in the field up to the present day.
 
The next talks are:
16.1. at 9:15 in Kumpula Exactum D123: Arto Klami "Probabilistic programming: Bayesian modeling made easy"
23.1. at 9:15 in Otaniemi CS building T5: Juho Rousu "Metabolite identification through machine learning"
30.1. at 9:15 in Kumpula Exactum D123: Jukka Corander "Likelihood-free inference and predictions for computational epidemiology”

Combinatorial Algorithms with Applications in Learning Graphical Models

Lecturer : 
Juho-Kustaa Kangas
Event type: 
Doctoral dissertation
Doctoral dissertation
Respondent: 
Juho-Kustaa Kangas
Opponent: 
James Cussens, Senior Lecturer, University of York
Custos: 
Professor Petri Myllymäki, University of Helsinki
Event time: 
2016-12-09 14:00 to 17:00
Place: 
Exactum, Auditorium CK112, Gustaf Hällströmin katu 2b, Helsinki
Description: 

Abstract

Graphical models are a framework for representing joint distributions over random variables. By capturing the structure of conditional independencies between the variables, a graphical model can express the distribution in a concise factored form that is often efficient to store and reason about.

As constructing graphical models by hand is often infeasible, a lot of work has been devoted to learning them automatically from observational data. Of particular interest is the so-called structure learning problem, of finding a graph that encodes the structure of probabilistic dependencies. Once the learner has decided what constitutes a good fit to the data, the task of finding optimal structures typically involves solving an NP-hard problem of combinatorial optimization. While first algorithms for structure learning thus
resorted to local search, there has been a growing interest in solving the problem to a global optimum. Indeed, during the past decade multiple exact algorithms have been proposed that are guaranteed to find optimal structures for the family of Bayesian networks, while first steps have been taken for the family of decomposable graphical models.

This thesis presents combinatorial algorithms and analytical results with applications in the structure learning problem. For decomposable models, we present exact algorithms for the so-called full Bayesian approach, which involves not only finding individual structures of good fit but also computing posterior expectations of graph features, either by exact computation or via Monte Carlo methods.

For Bayesian networks, we study the empirical hardness of the structure learning problem, with the aim of being able to predict the running time of various structure learning algorithms on a given problem instance. As a result, we obtain a hybrid algorithm that effectively combines the best-case performance of multiple existing techniques.

Lastly, we study two combinatorial problems of wider interest with relevance in structure learning. First, we present algorithms for counting linear extensions of partially ordered sets, which is required to correct bias in MCMC methods for sampling Bayesian network structures. Second, we give results in the extremal combinatorics of connected vertex sets, whose number bounds the running time of certain algorithms for structure learning and various other problems.

CS Forum: Roger Wattenhofer

Lecturer : 
Roger Wattenhofer
Event type: 
Guest lecture
Event time: 
2016-11-18 14:15 to 15:00
Place: 
T2, CS building, Konemiehentie 2
Description: 

 

Speaker: Prof Roger Wattenhofer
Speaker affiliation: ETH Zurich
Host: Prof Jukka Suomela
Time: 14:15 (coffee at 14:00)
Venue: T2, CS building, Konemiehentie 2

 

 

Cryptocurrencies: Bitcoin, Blockchain & Beyond

Abstract

I will first give a short introduction to the Bitcoin system, explaining some of the basics such as transactions and the blockchain. Then, I will discuss some interesting technical aspects in more detail, regarding the stability, security, and scalability of Bitcoin. In particular, I will discuss Bitcoin's eventual consistency, and the related problem of double spending. I will shed some light into our findings regarding the bankruptcy of MtGox, previously the dominant Bitcoin exchange service. And I will present duplex micropayment channels. Apart from scalability, these channels also guarantee end-to-end security and instant transfers, laying the foundation of a network of payment service providers.

Bio

Roger Wattenhofer is a full professor at the Information Technology and Electrical Engineering Department, ETH Zurich, Switzerland. He received his doctorate in Computer Science in 1998 from
ETH Zurich. From 1999 to 2001 he was in the USA, first at Brown University in Providence, RI, then
at Microsoft Research in Redmond, WA. He then returned to ETH Zurich, originally as an assistant
professor at the Computer Science Department.

Roger Wattenhofer's research interests are a variety of algorithmic and systems aspects in computer science and information technology, currently in particular wireless networks, wide area networks, mobile systems, social networks, and physical algorithms. He publishes in different communities: distributed computing (e.g., PODC, SPAA, DISC), networking (e.g., SIGCOMM, MobiCom, SenSys), or theory (e.g., STOC, FOCS, SODA, ICALP). He recently published the book "The Science of the Blockchain". A complete CV is available here.

Pages