Multi-user resource-sharing problem for the Internet

Event type: 
Doctoral dissertation
Doctoral dissertation
Respondent: 
Andrey Lukyanenko
Opponent: 
Professor Leon Petorsjan
Custos: 
Professor Jussi Kangasharju
Event time: 
2010-11-17 12:00 to 14:00
Place: 
University of Helsinki, Prothania campus (Yliopistonkatu 3), PIII auditorium
Description: 

Andrey Lukyanenko will defend his thesis "Multi-user resource-sharing problem for the Internet" on 17 November at 12:00 in Porthanian, auditorium PIII. His opponent is Professor Leon Petorsjan and custos Professor Jussi Kangasharju.  The thesis is available at the link above.

Popular abstract:

In this thesis we study a series of multi-user resource-sharing problems for the Internet, which involve distribution of a common resource among participants of multi-user systems (servers or networks). We study concurrently accessible resources, which for end-users may be exclusively accessible or non-exclusively. For all kinds we suggest a separate algorithm or a modification of common reputation scheme. Every algorithm or method is studied from different perspectives: optimality of protocols, selfishness of end users, fairness of the protocol for end users. On the one hand the multifaceted analysis allows us to select the most suited protocols among a set of various available ones based on trade-offs of optima criteria. On the other hand, the future Internet predictions dictate new rules for the optimality we should take into account and new properties of the networks that cannot be neglected anymore.

Detection of Anomalies in Time Series from Different Contexts

Lecturer : 
Antonio Neme
Event type: 
HIIT seminar
Event time: 
2010-10-25 13:15 to 14:45
Place: 
Otaniemi, T-talo, Computer Science building, hall T2
Description: 

Our next speaker for HIIT Otaniemi seminar series is Antonio Neme from the "Complex Systems" group of the National Autonomous University of Mexico.

All ICS@Aalto researchers are also warmly welcome to attend the seminar!

HIIT seminar Otaniemi, Monday October 25, 13:15
Location: Computer Science building, hall T2

Short Bio:
Antonio Neme received his Ph.D. degree in computer science from the National Autonomous University of Mexico. He works a researcher at the "Complex Systems" group of the National Autonomous University of Mexico since 2005. He is a visiting researcher in the Department of Information and Computer Science of the Aalto University School of Science and Technology. His research interests include data mining, unsupervised learning, agent-based modeling of social phenomena, and biomathematics.

Title:
Detection of Anomalies in Time Series from Different Contexts

Abstract:
Anomaly detection is an important task in several contexts, including weather conditions, environmental dynamics, biological sequences, and even in the arts. A system able to detect anomalies is presented with instances of normal or habitual behavior, generally in terms of time series, and it has to be able to learn relations and features that characterize that single class. Those relations and features allow the system to detect situations previously unseen. In this seminar, a methodology based in information theory and unsupervised learning is presented. Results covering different contexts are presented. First, in the arts, we present the case of authorship identification for several texts, and we also present the analysis over the writings from Iris Murdoch in both, the novels written in the pre-illness period, and the last novel written with mental handicaps. In the bioinformatics context, we present preliminary results in which our working methodology is able to detect external sequences to a given organism, which is related to horizontal gene transfer detection.
 

Structured prior information in probabilistic machine learning with applications in gene regulation modelling

Lecturer : 
Antti Honkela
Event type: 
HIIT seminar
Event time: 
2010-10-22 10:15 to 11:45
Place: 
Kumpula, Exactum C222
Description: 

Talk announcement:
HIIT Seminar Kumpula, Friday Oct 22, 10:15 a.m., Exactum C222

The next HIIT Seminar Kumpula presenter is Dr. Antti Honkela.
Antti was recently appointed to the position of senior researcher at HIIT.

The talk will include a general introduction.

Cookies are served - welcome!

--Matti Järvisalo

P.S. Please also note the updated HIIT Seminar Kumpula schedule below.

-------------------------------------------------------------------

TITLE:
Structured prior information in probabilistic machine learning with
applications in gene regulation modelling

SPEAKER:
Antti Honkela
Helsinki Institute for Information Technology

ABSTRACT:
Bridging the gap between purely data-driven methods and more
traditional computational modelling with physically based models is an important challenge in machine learning. We address this by combining a differential equation model of gene regulation and a probabilistic Gaussian process model. The resulting model captures the relevant degrees of freedom in the process very effectively and allows making very accurate predictions of potential targets of regulator genes (transcription factors).

This is joint work with Neil D. Lawrence, Magnus Rattray and Michalis Titsias.

BIO:
Dr. Antti Honkela is a docent of statistical machine learning at the
Department of Information and Computer Science at Aalto University. He received his PhD in Computer and Information Science at Helsinki University of Technology (TKK) in 2005. Since then he has been a postdoc at TKK (more recently Aalto University) at the Centre of Excellence in Adaptive Informatics Research and under an Academy of Finland Postdoctoral researcher's project. He has also spent 12 months at the University of Manchester, UK. His research interests include Bayesian machine learning and approximate inference as well as computational systems biology.

 

Data Aggregation: Balancing Delay and Communication Costs

Lecturer : 
Lauri Ahlroth
Event type: 
HIIT seminar
Event time: 
2010-10-18 13:15 to 14:45
Place: 
Otaniemi, Computer Science Building, Hall T1
Description: 

Our next speaker for HIIT Otaniemi seminar series is Lauri Ahlroth from the "Combinatorial Algorithms and Computation" group. Before this talk, Prof. Pekka Orponen will give a short overview on the research areas of the group.

All ICS@Aalto researchers are also warmly welcome to attend the seminar!

HIIT seminar Otaniemi, Monday October 18, 13:15
Location: Computer Science building, hall T1 (!!! Please note that seminar has been moved to hall T1 for this week. !!!)

Lauri Ahlroth
Helsinki Institute for Information Technology HIIT
Department of Information and Computer Science
Aalto University School of Science and Technology

Title:
Data Aggregation: Balancing Delay and Communication Costs

Abstract:
Making a choice between waiting and taking action is a fundamental and often a nontrivial problem. Action is usually preceded by initialization costs, which suggests postponing action and combining many similar events together. On the other hand, there are usually delay costs attached to postponing action. Both of these two costs cannot be minimized simultaneously, and minimizing the sum leads to difficult combinatorial problems.

In this talk, we introduce a data aggregation model with a hierarchical structure on the initialization costs. We also present new results related to our recently submitted manuscript 'Centralized Algorithms for Data Aggregation in Tree Networks'.

 

Parsimonious Modelling

Lecturer : 
Jaakko Hollmén
Event type: 
HIIT seminar
Event time: 
2010-10-11 13:15 to 15:46
Place: 
Computer Science building, hall T2
Description: 

As you already know, HIIT has seminar sessions also on Mondays. These sessions are going through the different research groups of HIIT hosted in  Otaniemi. Our next speaker is Jaakko Hollmén from "Parsimonious Modelling" group.

All ICS@Aalto researchers are also warmly welcome to attend the seminar!

HIIT seminar Otaniemi, Monday October 11, 13:15
Location: Computer Science building, hall T2

Jaakko Hollmén
Helsinki Institute for Information Technology HIIT
Department of Information and Computer Science
Aalto University School of Science and Technology

Title:
Parsimonious Modelling

Abstract:
The research group Parsimonious Modelling develops computational methods for data analysis and applies these methods on two particular application fields: cancer genomics and environmental informatics. Both of these application fields exhibit problems of high dimensional data and complex, unknown interactions between measurements.

Parsimonious modeling aims at achieving maximally simple or compact models as a result of the data analysis process. In practical problems, parsimony makes results more understandable and interpretable. For instance, feature variable selection aims at parsimony in terms of the number of variables in the model.

In the presentation, the research group will be presented; both methodological and applied research results will be reviewed.

Group homepage: http://hiit.fi/pm

Pages