Previous seminars

A. Badie Modiri - Efficient limited time reachability estimation in temporal networks

Monday 24.06.2019 13.30 - 14.00

Seminar room A142, T-building

Efficient limited time reachability estimation in temporal networks

Time-limited states characterise several dynamical processes evolving on the top of networks: During epidemic spreading infected agents may recover after some times, or in travel routing systems passengers may not wait forever for a connection. These systems can be commonly described as limited waiting-time processes. As an efficient representation of temporal networks, temporal event graphs have been proposed recently, which could map time-respecting paths all at once to study how they form connected structures in the temporal network fabric. However, their analysis has been limited to their weakly connected components, which only give an upper bound for in- and out-components determining the downstream outcome of any dynamical processes. Here we propose a probabilistic counting algorithm, which gives estimates of the in- or out-reachability (with any chosen waiting-time limit) for every starting event in a temporal network. The computational complexity of this method allows measurements for temporal networks with hundreds of millions of events with current commodity computing hardware. This opens up the possibility to analyse reachability, spreading processes, and other dynamics in large temporal networks in completely new ways; to compute centralities based on global reachability for all events; or to find the exact node and time, which could lead to the largest epidemic outbreak with high probability.


M. Asllani - Hopping in the crowd to unveil network topology

Malbor Asllani (Namur, Oxford)

Thursday 2018-07-12 14.00 - 15.00

Seminar room T5, T-building

Hopping in the crowd to unveil network topology

We introduce a nonlinear operator to model diffusion on a complex undirected network under crowded conditions. We show that the asymptotic distribution of diffusing agents is a nonlinear function of the nodes’degree and saturates to a constant value for sufficiently large connectivities, at variance with standard diffusion in the absence of excluded-volume effects. Building on this observation, we define and solve an inverse problem, aimed at reconstructing the a priori unknown connectivity distribution. The method gathers all the necessary information by repeating a limited number of independent measurements of the asymptotic density at a single node, which can be chosen randomly. The technique is successfully tested against both synthetic and real data and is also shown to estimate with great accuracy the total number of nodes.

Reference: M. Asllani, T. Carletti, F. Di Patti, D. Fanelli, F. Piazza, Phys. Rev. Lett. 120, 158301 (2018)

P. Mapar - Machine Learning for Enzyme Promiscuity

Parisa Mapar (Aalto)

Monday 2018-06-18 14.00 - 15.00

Seminar room T5, T-building

Machine Learning for Enzyme Promiscuity

With the discovery of an increasing number of catalytically promiscuous enzymes, which are capable of catalyzing multiple reactions, the traditional view of enzymes as highly specific proteins has been brought into question. The significant implications of protein promiscuity for the theory of enzyme evolution suggest that this inherent feature can be utilized as the seed for engineering new functions in biotechnology and synthetic biology as well as in drug design. Therefore, understanding protein promiscuity is becoming even more important as it provides new insights into the evolutionary process that has led to such vast functional diversity. While there have been numerous efforts devoted to recognizing the determinants of promiscuity, till date, this pertinent question regarding the distinctions between specialized enzymes and promiscuous enzymes has remained unanswered.

As an in silico approach, in this thesis, we attempt to find a predictive model which can accurately classify unseen proteins into catalytically promiscuous and non-promiscuous. To this end, we exploit different representations and properties of proteins, and adopt different computational approaches accordingly. The role of proteins sequences as indicators of promiscuity is investigated by means of the BLAST algorithm as well as string kernels. Additionally, to validate the interplay between proteins’ three-dimensional structures and their promiscuous behaviors, we employ a novel method which is modeling the topological details of proteins as graphs. Graph kernel functions are then applied to measure the structural similarities between the 3D structures of proteins. The classification is performed using SVM as a kernel-based method. The results indicate that proteins’ sequences have limited bearings on promiscuity. Conversely, proteins’ 3D structures can reliably predict whether a protein has promiscuous activities with an accuracy of 96%. Our best results are achieved using the Weisfeiler-Lehman subtree graph kernel and the secondary structure information of proteins.

J. Ranta - Sleep classification in infants using a bed mattress sensor and analysis of ECG

Jukka Ranta

Thursday 2018-05-31 14.00 - 14.30

Seminar room B337, T-building

Sleep classification in infants using a bed mattress sensor and analysis of ECG

Infants spend the majority of their time asleep. Although extensive studies have been carried out, the role of sleep for infant cognitive, psychomotor, temperament and developmental outcomes is not clear. The current contradictory results may be due to the limited precision when monitoring infant sleep for prolonged periods of time, from weeks to even months. Sleep-wake cycle can be assessed with sleep questionnaires and actigraphy, but they cannot separate sleep stages. The gold standard for sleep state annotation is polysomnography (PSG), which consist of several signal modalities such as electroencephalogram, electrooculogram, electrocardiogram (ECG), electromyogram, respiration sensor and pulse oximetry. A sleep clinician manually assigns sleep stages for 30 sec epochs based on the visual observation of these signals. Because method is obtrusive and laborious it is not suitable for monitoring long periods. There is, therefore, a need for an automatic and unobtrusive sleep staging approach.

In this work, a set of classifiers for infant sleep staging was created and evaluated. The cardiorespiratory and gross body movement signals were used as an input. The different classifiers aim to distinguish between two or more different sleep states. The classifiers were built on a clinical sleep polysomnography data set of 48 infants with ages ranging from 1 week to 18 weeks old (a median of 5 weeks). Respiration and gross body movements were observed using an electromechanical film bed mattress sensor manufactured by Emfit Ltd. ECG of the PSG setup was used for extracting cardiac activity. Signals were preprocessed to remove artefacts and an extensive set of features (N=81) were extracted on which the classifiers were trained.

The NREM3 vs other states classifier provided the most accurate results. The median accuracy was 0.822 (IQR:0.724-0.914). This is comparable to previously published studies on other sleep classifiers, as well as to the level of clinical interrater agreement. Classification methods were confounded by the lack of muscle atonia and amount of gross body movements in REM sleep.

The proposed method could be readily applied for home monitoring, as well as for monitoring in neonatal intensive care units.

E. Ryyppö - Internal connectivity and topological roles of nodes in functional brain networks

Elisa Ryyppö (Aalto)

Wednesday 2018-05-30 14.00 - 14.30

Seminar room A346, T-building

Internal connectivity and topological roles of nodes in functional brain networks

Many real-life phenomena consist of a number of interacting elements and can thus be modeled as a complex network. The human brain is an example of such a system where the neuronal information processing of the brain is characterized by interaction and information change between different brain regions.

In this thesis, we examine functional brain networks estimated from functional magnetic resonance imaging (fMRI) data. In the analysis the small measurement units, voxels, are grouped into larger entities that represent supposedly functionally homogeneous brain regions referred to as Regions of Interest (ROIs). Despite their assumed homogeneity, it has been demonstrated that the voxels within a ROI exhibit spatially and temporally varying correlation structure. This gives rise to a concept referred to as internal connectivity.

On the larger scale, the ROIs form a brain network where each ROI has its role in the structure of the network topology, i.e. a topological role. Topological roles have been suggested to be indicative of the node's functional specialization. On the other hand, it has been argued that internal connectivity may relate to the mechanisms the ROI uses to interact with its neighbors in the functional brain network. This thesis combines these two ideas. To this end, we aim to predict the ROI's topological role from its internal connectivity features. We find that using internal connectivity features as model variables increases the classification accuracy in comparison to a baseline classifier.

These results suggest that there is a relationship between internal connectivity and the ROI's topological role. This link provides a basis for faster and more computationally efficient topological role estimation. Further, it helps to better understand the mechanisms brain regions use to interact with each other. Both of these factors importantly increase our knowledge on brain function under different tasks and circumstances.

A. Badie Modiri — Error and Attack Tolerance of Public Transportation Networks: A Temporal Networks Approach

Arash Badie Modiri (Aalto)

Monday 2018-04-23 14.00 - 14.30

Seminar room A346, T-building

Error and Attack Tolerance in Public Transportation Networks: A Temporal Networks Approach

The behaviour of complex networks under attack provides insight into their internal structure. Furthermore, advances in methods for analysing temporal networks have enabled us to perform more detailed modelling of a certain subset of dynamic complex systems specially since frequency of events and temporal correlations play a role in dynamics of the system. In this report, the temporal network approach for study of robustness is applied to public transportation networks. The focus is on providing a set of tools to model different scenarios of attack and random failure, and processing the results with or without taking into account the origin-destination demand matrix frequently used in transportation network studies.

The results of the robustness analysis on temporal representation of public transport networks illustrate the distribution of accessibility and travel time after an attack or error and how it changes when more routes are removed. Furthermore we see that two methods of attack, one based on temporal betweenness centrality and one based on nominal capacity of routes, have a higher effect on increasing delays while attack methods based on centrality of routes in a static aggregated network do not perform any better than randomly removing routes.

A. Triana - Effects of spatial smoothing on group-level differences in the structure of functional brain networks

Ana Triana (Aalto)

Tuesday 2018-04-10 12.30 - 13.00

Seminar room A346, T-building

Effects of spatial smoothing on group-level differences in the structure of functional brain networks

Brain connectivity with functional Magnetic Resonance Imaging (fMRI) is a popular approach to investigate psychiatric disorders and to detect differences between healthy and clinical populations. In such studies, fMRI time-series must undergo several preprocessing steps to control for artifacts --such as head motion and breathing-- to increase the signal-to-noise ratio. Functional networks are created only after preprocessing, thus, choices on these steps may affect the end result. For example, it has been demonstrated that spatial smoothing induces changes in the functional network structure, but its effects on group-level differences in functional brain network structure are still unknown.

Here, we study such effects on the difference between Autism Spectrum Disorder (ASD) patients and healthy controls (TC) using resting-state data from the ABIDE initiative (N=33 matched subject pairs). We apply 4 different levels of spatial smoothing (0, 6, 8, 12mm) and compute functional networks using nodes defined in each subject’s native space. This allows us to reduce possible bias due to the implicit spatial smoothing used at the registration of the individual data to a common brain template.

We find that link weights are affected by smoothing. These effects are diverse, non-trivial and hard to predict, as they affect links differently. In particular, for full weight matrices, the weight distributions of both groups tend to become more similar for the links showing the most significant between-group differences (T-statistic>3). However, for certain links, the effect is the opposite. For links with smaller T-statistics, the effects appear random.

This has important consequences for group comparisons of functional networks: spatial smoothing distorts network differences between groups. Hence, we conclude that spatial smoothing should not be used in the fMRI preprocessing for group-level differences analysis using functional networks.

S. Heydari - Multichannel Social Signatures and Persistent Features of Ego Networks

Sara Heydari (Aalto)

Thursday 2018-03-15 9.00 - 9.45

Meeting room A346, T-building

Multichannel Social Signatures and Persistent Features of Ego Networks

The structure of egocentric networks reflects the way people balance their need for strong, emotionally intense relationships and a diversity of weaker ties. Egocentric network structure can be quantified with 'social signatures', which describe how people distribute their communication effort across the members (alters) of their personal networks.

Social signatures based on call data have indicated that people mostly communicate with a few close alters; they also have persistent, distinct signatures. To examine if these results hold for other channels of communication, here we compare social signatures built from call and text message data, and develop a way of constructing mixed social signatures using both channels. We observe that all types of signatures display persistent individual differences that remain stable despite the turnover in individual alters. We also show that call, text, and mixed signatures resemble one another both at the population level and at the level of individuals. The consistency of social signatures across individuals for different channels of communication is surprising because the choice of channel appears to be alter-specific with no clear overall pattern, and ego networks constructed from calls and texts overlap only partially in terms of alters. These results demonstrate individuals vary in how they allocate their communication effort across their personal networks and this variation is persistent over time and across different channels of communication.

S. Sallmen - Graphlets in Multilayer Networks

Sallamari Sallmen (Aalto)

Friday 2018-01-26 14.15 - 15.00

Seminar room T5, T-building

Graphlets in Multilayer Networks

Networks, consisting of nodes and edges, are used to model and analyze a wide array of phenomena in various disciplines. Recently, increasing attention has been given to more realistic network representations such as multilayer networks, which can include different types of interactions between nodes or other additional structural information. However, generalizing concepts devised for ordinary graphs to multilayer networks is in its early stages. In this thesis, the concept of graphlets and algorithms utilizing them to analyze networks are generalized to multilayer networks. Graphlet analysis has been applied to alignment-free network comparison, and here such a comparison method is developed for multiplex networks, which are a type of multilayer networks. The ability of multilayer-graphlet-based distance measures to cluster similar networks is assessed by applying the measures for both real world and random networks, and the performance of these new multilayer methods are compared to methods based on ordinary networks. The results indicate that one can benefit from the multiplex measures when there exists clear multiplex structure in the network.

N. Tran Quang - When is Network Lasso Accurate: The Vector Case

Nguyen Tran Quang (Aalto)

Thursday 2017-11-23 13.15 - 14.00

Seminar room 1021-1022, TUAS-building

When is Network Lasso Accurate: The Vector Case

A recently proposed learning algorithm for massive network-structured data sets (big data over networks) is the network Lasso (nLasso), which extends the well-known Lasso estimator from sparse models to network-structured datasets. Efficient implementations of the nLasso have been presented using modern convex optimization methods. In this paper, we provide sufficient conditions on the network structure and available label information such that nLasso accurately learns a vector-valued graph signal (representing label information) from the information provided by the labels of a few data points.

Journal club: The price of complexity in financial networks

October 26, 2017 @ 14.15 - 15.00
Seminar room 1021-1022, TUAS-building
Presenter: Alexander Gurevich

Financial institutions form multilayer networks by engaging in contracts with each other and by holding exposures to common assets. As a result, the default probability of one institution depends on the default probability of all of the other institutions in the network. Here, we show how small errors on the knowledge of the network of contracts can lead to large errors in the probability of systemic defaults. From the point of view of financial regulators, our findings show that the complexity of financial networks may decrease the ability to mitigate systemic risk, and thus it may increase the social cost of financial crises.

Link to the article:

The price of complexity in financial networks
Authors: S. Battiston, G. Caldarelli, R. M. May, T. Roukny, J. E. Stiglitz

C. Gershenson - Improving Urban Mobility with Self-organization

C. Gershenson (IIMAS, UNAM, Mexico)

Thursday 2017-11-02 12.00 - 12.30

Seminar room T5, T-building

Improving Urban Mobility with Self-organization

Urban mobility is non-stationary, i.e. the precise number of vehicles or passengers is changing constantly, and with a limited predictability. Transportation systems will be more effective if they can adapt at the same timescales at which the demand changes. Self-organization offers one way of implementing this desired adaptability. I will present two examples: self-organizing traffic lights which achieve quasi-optimal performance and self-organizing public transportation systems which achieve supraoptimal performance.

Carlos Gershenson is a tenured, full time research professor at the computer science department of the Instituto de Investigaciones en Matemáticas Aplicadas y en Sistemas at the Universidad Nacional Autónoma de México (UNAM), where he leads the Self-organizing Systems Lab. He is also associated with the Centro de Ciencias de la Complejidad at UNAM, MIT’s Senseable City Lab, and ITMO University. He has a wide variety of academic interests, including complex systems, self- organization, urbanism, artificial life, evolution, cognition, artificial societies, and philosophy. More info at

C. Pineda - The random nature of rank dynamics

C. Pineda (IF, UNAM, Mexico)

Thursday 2017-11-02 12.30 - 13.00

Seminar room T5, T-building

The random nature of rank dynamics

Any set can be ranked by comparing a common property, such as size, age, or wealth. Ranks indicate who does one object compare to others of the same set. People have analyzed the rank distribution of words, cities, earthquakes, and networks, just to name a few. Rank distributions seem prevalent because they are general descriptions of diverse phenomena. As such, they have applications in many areas, from science to business. How does rank change in time? To explore this question, we have proposed the measure "rank diversity". Assuming that elements change their rank in time, the elements’ trajectories can be tracked in rank space. Rank diversity is the normalized number of different elements that appear at a specific rank at different times.

We have measured the rank diversity of a broad range of phenomena: languages, sports, earthquakes, economic systems, transportation systems, and social systems. We have found two different universality classes of rank diversity curves: for open systems (where elements enter and leave the ranking in time), rank diversity increases as a sigmoid with rank. The second class is for closed systems (where most elements do not leave or enter the ranking during the evolution); the diversity behaves as a semicircle. If rank diversity is so similar for different phenomena, and considering that the mechanisms to determine rank change in every system might differ, what are the minimal assumptions required for reproducing the two classes of rank diversity? To answer this, we present a single null model, for both classes.

In this model an element from a list is picked at random and replaced at a new random position. The solutions have a drift component that obeys a leaking diffusion-like equation with quadratic coefficients, and a Levy-type component that increases in size linearly with time. Its predictions show that a good portion of the data analyzed can be explained with it. Important quantities such as the first step probability can be accurately described with such a model, both in the open and closed situations.

Carlos Pineda is a tenured, full-time professor at the Physics Institute of Universidad Nacional Autónoma de México (UNAM), and is also associated with the Centro de Ciencias de la Complejidad at UNAM. His interests include quantum information, many-body problems, and complex systems.

O. Korhonen - Regions of Interest as nodes of dynamic functional brain networks

Onerva Korhonen (Aalto)

Thursday 2017-11-09 12.15 - 13.00

Seminar room T5, T-building

Regions of Interest as nodes of dynamic functional brain networks

The properties of functional brain networks strongly depend on how their nodes are chosen. Commonly, nodes are defined by Regions of Interest (ROIs), pre-determined groupings of fMRI measurement voxels. Earlier, we have demonstrated that the functional homogeneity of ROIs, captured by their spatial consistency, varies widely across ROIS in commonly-used brain atlases. Here, we ask how ROIs behave as nodes of dynamic brain networks. To this end, we use two measures: spatiotemporal consistency measures changes in spatial consistency across time and network turnover quantifies the changes in the local network structure around a ROI. We find that spatial consistency varies non-uniformly in space and time, which is reflected in the variation of spatiotemporal consistency across ROIs. Further, we see time-dependent changes in the ROIs' network neighborhoods, resulting in high network turnover. This turnover is nonuniformly distributed across ROIs: ROIs with high spatiotemporal consistency have low network turnover. Finally, we reveal rich internal, voxel-level correlation structure inside ROIs.Because the internal structure and connectivity of ROIs vary in time, the common approach of using static node definitions may be surprisingly inaccurate. Therefore, network neuroscience would greatly benefit from node definition strategies tailored for dynamical networks.

Article in arXiv:

J. Török - Age prediction using egocentric network

János Török (BME)

Thursday 2017-07-27 14.15 - 15.00

Lecture room AS3, TUAS-building

Age prediction using egocentric network

Prediction is often done by supervised learning where the many parameter model (neural network) is trained by the some part of data, and the accuracy of the prediction is then verified on the rest. The problem with these approaches that these systems often behave as a black box and not much can be learned from the results. Here we present a simple model which has basically no adjustable parameter but can predict the age of an ego with 85-90% success rate from the age of its acquaintances, using the egocentric network.

J. Török - Cascading collapse of online social networks

János Török (BME)

Thursday 2017-07-18 14.15 - 15.00

Lecture room AS3, TUAS-building

Cascading collapse of online social networks

Online social networks have increasing influence on our society, they may play decisive roles in politics and can be crucial for the fate of companies. Such services compete with each other and some may even break down rapidly. Using social network datasets we show the main factors leading to such a dramatic collapse. At early stage mostly the loosely bound users disappear, later collective effects play the main role leading to cascading failures. We present a theory based on a generalised threshold model to explain the findings and show how the collapse time can be estimated. Our results shed light to possible mechanisms of instabilities in other competing social processes.

T. Shimada - SOC and non-SOC behaviors in simple models of evolving systems

Takashi Shimada (University of Tokyo)

Thursday 2017-07-20 14.15 - 14.45

Lecture room AS3, TUAS-building

SOC and non-SOC behaviors in simple models of evolving systems

For understanding of evolving systems, the robustness is one of the most essential property. The self-organized-criticality, which argue the system's nature to evolve toward a critical state, is a famous example.
Recently a new type of mechanism of governing the robustness of evolving open systems was found using a simple graph dynamics model (EOS model). However, the EOS model does not show any critical property. To fill the gap between these two distinct classes of mathematical processes, here we consider the EOS model with an SOC-model-like node deletion rule: node with minimum fitness is replaced by a new node and hence the system size is kept constant. It will be shown that the modified EOS model shows critical and non-critical behavior depending on the number of interactions per element.

F. Ogushi - Bidirectional interaction enhances the robustness of the system

Fumiko Ogushi (Tohoku University)

Thursday 2017-05-04 14.45 - 15.15

Lecture room AS3, TUAS-building

Bidirectional interaction enhances the robustness of the system

An essential feature of complex systems like ecological, biological and social systems is that they are open. These systems can exist with inclusion of new elements and disappearance of old elements. On the other hand, in real systems, the interactions can be considered to unidirectional and bidirectional. Our question is how the bidirectionality of interactions influences on growth of the real systems. We investigated the effect of bidirectional interactions on the robustness of evolving open system using a simple model [1]. We found that the system with purely bidirectional interactions can grow with two-fold average degree, in comparison with the system with purely unidirectional interactions [2]. This shift of the transition point comes from the reinforcement of each element, not from a change in structure of the emergent system.

[1]T. Shimada, “A universal transition in the robustness of evolving open system”, Scientific Reports 4, 4082 (2014)
[2] F. Ogushi, J. Kertesz, K. Kaski, and T. Shimada, “Enhanced robustness of evolving open systems by the bidirectionality of interactions between elements”, arXiv: 1703.04383v1 (2017)

H.-H. Jo - Configuration models for correlated bursty dynamics

Hang-Hyun Jo (APCTP)

Tuesday 2017-08-01 14.15 - 15.00

Lecture room AS3, TUAS-building

Configuration models for correlated bursty dynamics


J. Kertész - The hybrid percolation transition

János Kertész (CEU, BME, Aalto)

Tuesday 2017-07-25 14.15 - 15.00

Lecture room AS3, TUAS-building

The hybrid percolation transition

Percolation is a paradigmatic example of second order phase transitions where the order parameter changes continuously (in contrast to first order transitions). However, there are dynamic versions of the percolation model, where the order parameter has a jump at the transition. Such models include the Watts cascade model, the interdependent network model, k-core percolation and the SWIR epidemic model. Interestingly, all these models show features of both first order and second order transitions at the same time: There is a discontinuity in the order parameter and there are scaling phehomena too. We show that there are two sets of exponents describing the transition in these cases: One characterizes the behavior of the order parameter, the other one that of the avalanches (or cascades). There is a scaling law connecting these exponents and one of them is given exactly for the case of infinite range interdependent networks. We show on the Erdős-Rényi graph that there is a universal mechanism behind the hybrid percolation transition: The process starts with the evolution of a cascade as a critical branching tree (hence the scaling properties) during which latent nodes accumulate. Due to the finitness of the samples loops occure which ignite some latent nodes leading rapidly to a global cascade and a jump in the order parameter. These phenomena separate in time, enabling intervention during the first period and giving an interpretation of the "golden time" in epidemiology. We show that the dependence of the length of the golden time on the size N of the system is ~N^1/3 if the process starts from an O(1) part of the system and it is ~N^1/4 if from an O(N) part.

Y. Murase - An open-source software framework for comprehensive simulations

Yohsuke Murase (University of Tokyo)

Thursday 2017-08-03 14.15 - 15.00

Lecture room AS3, TUAS-building

An open-source software framework for comprehensive simulations

In this talk, I’m going to give an overview on our project of developing a software framework for parameter-space exploration, named OACIS, which is useful to manage vast amount of simulation jobs and results in a systematic way. Recent development of high-performance computers enabled us to explore parameter spaces comprehensively, however, in such cases, manual management of the workflow is practically impossible. OACIS is developed aiming at reducing the cost of these repetitive tasks when conducting simulations by automating job submissions and data management. The source code is available at

A. Keurulainen - Deep neural networks applications for education

Antti Keurulainen (Aalto)

Thursday 2017-09-28 14.15 - 15.00

Seminar room 1021-1022, TUAS building

Deep neural networks applications for education

Antti Keurulainen has been a visiting researcher at Aalto since 2015 with a focus on deep neural networks and their applications for education and human learning. Keurulainen completed his licentiate studies and graduated in the spring of 2017. The thesis topic was “Applications of deep neural networks for assisting human learning”.

Educational applications offer opportunities to make use of advanced deep neural network methods. In this talk, Keurulainen will present three different applications and the technology behind them. A relatively simple multilayer perceptron network is constructed and trained to assess the sentiment of student feedback texts collected from Aalto students. A more complicated memory augmented network is presented to assess essays. The third example is using sequence modeling and recurrent neural networks to estimate the skills of a learner based on the previous interactions between the student and learning material. The main emphasis is on the augmenting deep neural network with an external memory component, which is currently a hot topic in machine learning research community.

Journal Club: The multi-layer approach in network neuroscience

June 1, 2017 @ 14.15 - 15.00
Room T3, T-building
Presenter: Onerva Korhonen

Modeling the human brain as a network has opened new insights on the structure and function of the brain. However, the human brain is not a single, unchangeable network. On the contrary, network neuroscientists should be able to handle a number of different brain networks: structural and functional networks, networks at different frequency bands, networks evolving in time, and even networks of different subjects. In this journal club talk, I will discuss the multilayer network approach applied in the context of the brain: why would it be used and what could it offer to neuroscience?

The presentation is based on the following articles:

Multilayer modeling and analysis of human brain networks
Author: De Domenico

Structure-function clustering in multiplex brain networks
Authors: Crofts, Forrester & O'Dea

Frequency-based brain networks: From a multiplex framework to a full multilayer description
Authors: Buldú & Porter

L. Leskelä - Clustering coefficients in large directed graphs

Lasse Leskelä (Aalto)

Thursday 2017-05-04 14.15 - 15.00

Room A136 (T6), T-building

Clustering coefficients in large directed graphs

I will discuss a notion of clustering in directed graphs which describes how likely two followers of a node are to follow a common target. The associated network motifs, called dicliques or bi-fans, have been found to be key structural components in various real-world networks. A two-mode statistical network model consisting of actors and auxiliary attributes is introduced, where an actor i decides to follow an actor j whenever i demands an attribute supplied by j. This directed random graph model admits nontrivial clustering properties of the aforementioned type, as well as power-law indegree and outdegree distributions. The talk is based on joint work with Mindaugas Bloznelis (U Vilnius), available at arXiv:1607.02278.

J. Autere - EIT Digital's Industrial Doctoral Program

Jussi Autere (Aalto)

Thursday 2017-04-06 14.00 - 15.00

Room A136 (T6), T-building

EIT Digital's Industrial Doctoral Program

EIT Digital a leading European digital innovation and entrepreneurial education organisation has started an industrial doctoral program. It offers doctoral student positions to researchers interested in subjects relevant to the industry.

A. Ghosh - Finnish migration patterns: Proximity with parents and siblings across the life course

Asim Ghosh (Aalto)

Thursday 2017-03-30 14.00 - 15.00

Room A136 (T6), T-building

Finnish migration patterns: Proximity with parents and siblings across the life course

In order to get insight into the migration patterns of people across their life course and how they are separating from their parents and siblings, we analyse their migration dynamics from one region (Maakunta) to another region in Finland. Finland is divided into 19 regions or Maakuntas and first we draw the flow of people moving from one region to another. We observe that the migration probability of people is higher at their early infancy due to their parents moving, but they also migrate with high probability at their early adulthood for their education or for jobs in different regions. We calculate the probability for staying with parents as a function of the age of children and find that when the childen's age is from 15 to 30 years they have a greater probability to move out from their parents' home. Comparing sons with daughters reveals that the daughters start to move out from their parents earlier and continue to do so with higher rate. We also study the cohabitation probability with siblings and half-siblings and find that the probability for two full sibling brothers to be in the same region is the highest in comparison with the other types of siblinghood pairs. Overall, we find that in Finnish society the females have larger probability to migrate than males do.

A. Jung - When is Network Lasso Accurate?

Alex Jung (Aalto)

Thursday 2017-04-13 14.00 - 15.00

Room A136 (T6), T-building

When is Network Lasso Accurate?

The network LASSO is a recently proposed method for clustering and optimization problems arising from massive network-structured datasets, i.e.,
for big data over networks. It is a variant of the well-known group LASSO which is underlying many methods in machine learning, statistics and signal processing involving sparsity assumptions. While much work has been devoted to studying efficient and scalable implementations, only little is known about conditions on the underlying network structure required by network Lasso to be accurate. In some of our most recent work we close this gap by giving precise conditions on the underlying network topology which guarantee the network lasso to be accurate.

Journal Club - Network Lasso: Clustering and Optimization in Large Graphs

March 9, 2017 @ 14.00 - 15.00
Room A136 (T6), T-building
Presenter: Jonathan Strahl

Authors: David Hallac, Jure Leskovec, Stephen Boyd

Convex optimization is an essential tool for modern data analysis, as it provides a framework to formulate and solve many problems in machine learning and data mining. However, general convex optimization solvers do not scale well, and scalable solvers are often specialized to only work on a narrow class of problems. Therefore, there is a need for simple, scalable algorithms that can solve many common optimization problems. In this paper, we introduce the network lasso, a generalization of the group lasso to a network setting that allows for simultaneous clustering and optimization on graphs. We develop an algorithm based on the Alternating Direction Method of Multipliers (ADMM) to solve this problem in a distributed and scalable manner, which allows for guaranteed global convergence even on large graphs. We also examine a non-convex extension of this approach. We then demonstrate that many types of problems can be expressed in our framework. We focus on three in particular — binary classification, predicting housing prices, and event detection in time series data — comparing the network lasso to baseline approaches and showing that it is both a fast and accurate method of solving large optimization problems.

Download the article here
Presentation slides

P. della Briotta Parolo - Scientific Snowballs

Pietro della Briotta Parolo (Aalto)

Monday 2016-12-12 11.00 - 12.00

Lecture hall AS3, TUAS building

Scientific Snowballs

Our idea is to analyze the spread of scientific ideas coming from a certain field/subfield in time following the chain of citations that follow the initial publications. The basic case study is based on initializing a set N of papers from a certain field/subfield/journal and year and assigning to each an initial scientific value,which can be initialized either equally (i.e. 1/N for each) or based on the citations each paper received (more value to more influential papers). Then we proceed to push the initial value through the network until the desired year is reached. This is done by organizing the network in topological order and looping through the ordered network, thus making sure the values are always pushed forward. Similarly to the initialization of value, the pushing can be performed either equally (each child paper receives an equal amount of value) or based on citations (each child papers gets value proportionally to its number of citations). At the end of the pushing, the value is gathered from paper level to field/subfield/journal level.

The initial results show that all fields follow a similar behavior, retaining most of their scientific value, but losing it quickly in the first years and more gradually as time goes by, eventually reaching a plateau. On the other hand the distribution of values within the field seems to keep the same distribution of value across it subfields. The next step is to quantify the rate of loss of value within a field and try to compare the value loss across fields and time.

J. Rousu - Predicting structured data

Juho Rousu (Aalto)

Monday 2016-12-05 11.00 - 12.00

Lecture hall AS3, TUAS building

Predicting structured data

In modern data analysis, data does not always come in neat tables of numbers, each column corresponding to a variable of interest. In particular, many data have internal structure, with statistical dependencies arising between close parts of the structure. In my talk I will introduce structured (output) prediction, a branch of machine learning that aims to take advantage of the data structures (e.g. sequences, trees, networks) and the embedded statistical dependencies, in order to make predictive models more accurate and inference more efficient. I will illustrate the methods by two examples: Network response prediction (Su et al, ICML 2014), that asks to predict a subnetwork that responds to a stimulus arising in a node of a larger network, and Metabolite identification (Brouard et al. 2016, ISMB 2016), where the outputs correspond to graphs representing small molecules.

K. Bhattacharya - Network of families resulting from unions in the FINNFAMILY data

Kunal Bhattacharya (Aalto)

Monday 2016-11-07 11.00 - 12.00

Lecture hall AS3, TUAS building

Network of families resulting from unions in the FINNFAMILY dat

The FINNFAMILY dataset is a multi-generational representative dataset from late 20th century Finland. The data is derived from the National Population Register of Finland and consists of 60,000 randomly selected Finns (index-persons) from six birth cohorts. The dataset allows to sample "unions" (male-female pair who are not known to be genetically related and whose union resulted in one or more offspring) between extended families. Analysis of these linkages will show that “social gravity” plays an important role in structuring the network between families. In addition, the analysis will also indicate the overall patterns of kinship in contemporary Finland.

S. Heydari - Social Signatures

Sara Heydari (Aalto)

Monday 2016-10-31 11.00 - 12.00

Lecture hall AS3, TUAS building

Social Signatures

Humans have limited time and resources for socializing. People do not distribute these limited resources evenly between all their acquaintance but they dedicate most of their communication effort to a few alters while distributing the rest between numerous alters. Although this disparity exists in communication patterns of all individuals, there are individual differences between patterns. Social signatures [1], which are communication patterns constructed from mobile phone records, can reflect both the general disparity in communication and individual differences in the ways how people prefer to shape their networks. In this talk we will first review past research works on social signatures, and then explore possible relationships between the shape of social signatures and characteristic attributes of individuals, networks attributes(degree, centrality) and homophily. We also talk about differences and similarities of social signatures constructed from communication data through different channels (phone calls and text messages).

[1] Saramäki, Jari, et al. "Persistence of social signatures in human communication." Proceedings of the National Academy of Sciences 111.3 (2014): 942-947.

D. Monsivais-Velazquez - Seasonal, geographical and thermal influence on human resting patterns inferred from mobile phone data

Daniel Monsivais-Velazquez (Aalto)

Monday 2016-10-24 11.00 - 12.00

Lecture hall AS3, TUAS building

Seasonal, geographical and thermal influence on human resting patterns inferred from mobile phone data

We study influence of seasonally and geographically related daily dynamics of daylight and daily temperature on human resting or sleeping patterns using mobile phone data of a very large number of individuals. We characterize the two observed daily "resting" or inactivity periods for calling in the aggregated pattern of mobile phone using population, inferring it to represent the sleeping times of people in cities. We find that the nocturnal resting period is strongly influenced by the length of the daylight, and that its seasonal variation depends on latitude, such that for people living in cities lying in two parallels separated by eight degrees, the difference of the resting period between summer and winter for southern cities is almost twice the value for the northern cities. We also observe that the duration of the resting period taking place in the afternoons is influenced by the temperature, and that there is a threshold from which this influence sets on. Finally, we argue that yearly dynamics of the afternoon and nocturnal resting periods are entangled, in such a way that the total daily resting period is constant along the year.

Article in arXiv:

D. Ferreira - Understanding human behaviour with technology: challenges, tools & methods

Denzil Ferreira (University of Oulu)

Monday 2016-11-01 12.00 - 13.00

Lecture hall AS3, TUAS building

Understanding human behaviour with technology: challenges, tools & methods

Mobile phones have an increasing spectrum of built-in sensors, such as motion, light, atmospheric pressure. These sensors are primarily used to enhance the user experience with the device, such as detecting the screen orientation. More important for scientists, these sensors offer the potential to sense and reason about the user’s environment, or in other words, the user’s context. Mobile phones are the most widespread personal sensing device and provide an exciting opportunity for wider cross-disciplinary research to attain a better understanding of human behaviour by analysing the users’ unique context.

Yet the biggest challenge in conducting user studies is the scientists’ need to build software and logging tools from scratch, often without proper development experience, over and over again. More critically, multidisciplinary research becomes increasingly challenging due to the diversity of applications and environments. Researchers have no infrastructure support for exchanging their expertise and to collaborate locally or remotely.

In this talk, we introduce AWARE, a tool that focuses on an infrastructure for sensing behavioural and social context from mobile phones sensors, to enable a better understanding of human and social behaviour, and to improve users’ understanding of their own quality of life. More importantly, it is a platform that supports reuse and sharing of mobile-based behavioural and social context and researchers’ expertise.

Ferreira’s research is on leveraging mobile technologies to improve people’s lives, understanding people’s frustrations and fix them! He develops and evaluate tools for better understanding how mobility and social context affects your wellbeing and others. He is a member of Center for Ubiquitous Computing ( research group at the University of Oulu, and also a member of Ubicomp Lab research group at the Human-Computer Interaction Institute at Carnegie Mellon University.

T. Alakörkkö - Monitoring daily behavioural patterns using mobile phone sensors and ballistocardiography for detection of mental health changes

Tuomas Alakörkkö (Aalto)

Monday 2016-10-17 11.00 - 12.00

Lecture hall AS3, TUAS building

Monitoring daily behavioural patterns using mobile phone sensors and ballistocardiography for detection of mental health changes

Increasing number of mental health problems in population have created a need for developing new patient monitoring methods for mental health care. Use of different type of sensors could automatize the monitoring which could help to provide mental health services for all in need with the limited resources. In these days, commonly used mobile phones have sensors which can be used to collect data related to human behaviour unobtrusively. These sensors can measure for example acceleration, ambient noise levels, brightness and usage of a mobile phone. Also, a ballistocardiographic sensor can be used to obtain information about rest which is known to be a factor in mental health.
A pilot study was conducted where mobile phone sensors and ballistocardiographic sensor were used to measure sixteen subjects for six weeks. Subjects also filled daily questionnaires related to mental health and sleep and wore an actigraph for two weeks which were used as a comparison for the passively collected data. This thesis presents and and evaluates methods for the analysis of the sensor data. These methods could be used in future studies to detect differences between mental health patients and normal population and to search for markers of mental health problems.

D. Hric - Network structure, metadata, and the prediction of missing nodes and annotations

Darko Hric (Aalto)

Monday 2016-09-12 11.00 - 12.00

Lecture hall AS3, TUAS building

Network structure, metadata, and the prediction of missing nodes and annotations

The empirical validation of community detection methods is often based on available annotations on the nodes that serve as putative indicators of the large-scale network structure. Most often, the suitability of the annotations as topological descriptors itself is not assessed, and without this it is not possible to ultimately distinguish between actual shortcomings of the community detection algorithms, on one hand, and the incompleteness, inaccuracy, or structured nature of the data annotations themselves, on the other. In this work we present a principled method to access both aspects simultaneously. We construct a joint generative model for the data and metadata, and a non-parametric Bayesian framework to infer its parameters from annotated data sets. We assess the quality of the metadata not according to its direct alignment with the network communities, but rather in its capacity to predict the placement of edges in the network. We also show how this feature can be used to predict the connections to missing nodes when only the metadata are available. By investigating a wide range of data sets, we show that while there are seldom exact agreements between metadata tokens and the inferred data groups, the metadata are often informative of the network structure nevertheless, and can improve the prediction of missing nodes. This shows that the method uncovers meaningful patterns in both the data and metadata, without requiring or expecting a perfect agreement between the two.

J. Török - Understanding and coping with extremism in an online collaborative environment

János Török (Budapest University of Technology)

Monday 2016-08-15 11.00 - 12.00

Lecture hall AS3, TUAS building

Understanding and coping with extremism in an online collaborative environment

The Internet has provided us with great opportunities for large scale collaborative public good projects. Wikipedia is a predominant example of such projects where conflicts emerge and get resolved through bottom-up mechanisms leading to the emergence of the largest encyclopedia in human history. Disaccord arises whenever editors with different opinions try to produce an article reflecting a consensual view. The debates are mainly heated by editors with extremist views. Using a model of common value production, we show that the consensus can only be reached if extremist groups can actively take part in the discussion and if their views are also represented in the common outcome, at least temporarily. We show that banning problematic editors mostly hinders the consensus as it delays discussion and thus the whole consensus building process. To validate the model, relevant quantities are measured both in simulations and Wikipedia which show satisfactory agreement. We also consider the role of direct communication between editors both in the model and in Wikipedia data (by analysing the Wikipedia talk pages). While the model suggests that in certain conditions there is an optimal rate of "talking" vs "editing", it correctly predicts that in the current settings of Wikipedia, more activity in talk pages is associated with more controversy.

H.-H. Jo - Burstiness parameter for finite event sequences

Hang-Hyun Jo (POSTECH/Aalto)

Monday 2016-08-08 11.00 - 12.00

Lecture hall AS3, TUAS building

Burstiness parameter for finite event sequences

Characterizing inhomogeneous temporal patterns in natural and social phenomena is important to understand underlying mechanisms behind such complex systems, hence even to predict and control them. Temporal inhomogeneities in event sequences have been described in terms of bursts that are rapidly occurring events in short time periods alternating with long inactive periods. The bursts can be quantified by a simple measure, called burstiness parameter, which was introduced by Goh and Barab\'asi [EPL \textbf{81}, 48002 (2008)]. The burstiness parameter has been widely used due to its simplicity, which however turns out to be strongly biased when the number of events in the time series is not large enough. As the finite size effects on burstiness parameter have been largely ignored, we analytically investigate the finite size effects of the burstiness parameter. Then we suggest an alternative definition of burstiness parameter that is unbiased and yet simple. Using our alternative burstiness parameter, one can distinguish the finite size effects from the intrinsic bursty properties in the time series. We also demonstrate the advantages of our burstiness parameter by analyzing empirical datasets.


T. Shimada - A transition in growth and robustness of evolving networks

Takashi Shimada (Department of Applied Physics, The University of Tokyo)

Monday 2016-08-01 11.00 - 12.00

Lecture hall AS3, TUAS building

A transition in growth and robustness of evolving networks

An important and universal feature of real complex systems, such as social, economic, engineering, ecological, and biological systems, is that those are open: in those systems, constituting elements are not fixed and the complexity emerges (at least persist) under successive introductions of new elements. Those systems sometimes grow, but also sometimes collapse. Therefore why and when, in general, we can have such open and complex systems is a fundamental question. In this talk, we revisit this classical problem using our very simple graph dynamics model. I will show that this model gives either continuous growth or stagnation in system size, depending on the model’s unique parameter: the average number of weighted links per node m. The system can grow only if the connection is moderately sparse, i.e. 5 m 18. We can further find that this transition originates from an essential balance of two effects: although having more interactions makes each node robust, it also increases the impact of the loss of a node [1, 2]. This novel relation might be a origin of the moderately sparse (average degree 10) network structure ubiquitously found in real world, and the non-trivial distribution function of the lifetime of elements [3].

[1] T. Shimada, in Springer monograph Mathematical Approaches to Biological Systems: networks, Oscillations and Collective Motions (Springer, 2015) p. 95-117.
[2] T. Shimada “ A universal transition in the robustness of evolving open systems ”, Scientific Reports Vol. 4, 4082 (2014).
[3] Y. Murase, T. Shimada, & Ito, N.“A simple model for skewed species-lifetime distributions ”, New J. of Physics 12, 063021 (2010).

Y. Murase - What does Big Data tell? Sampling the social network by communication channels

Yohsuke Murase (RIKEN Advanced Institute for Computational Science)

Wednesday 2016-08-10 11.00 - 12.00

Lecture hall AS3, TUAS building

What does Big Data tell? Sampling the social network by communication channels

Big Data has become the primary source of understanding the structure and dynamics of the society at large scale. The network of social interactions can be considered as a multiplex, where each layer corresponds to one communication channel and the aggregate of all them constitutes the entire social network. However, usually one has information only about one of the channels or even a part of it, which should be considered as a sample of the whole. Here we show by simulations and analytical methods that this sampling may lead to bias. For example, while it is expected that the degree distribution of the whole social network has a maximum at a value larger than one, we get with reasonable assumptions about the sampling method a monotonously decreasing distribution as observed in empirical studies of single channel data. We also find that assortativity may occur or get strengthened due to the sampling method. We analyze the far-reaching consequences of our findings.

Article:, J. Torok, Y. Murase, H.-H. Jo, J. Kertesz, K. Kaski "What does Big Data tell? Sampling the social network by communication channels".

G. Iñiguez - Global contagion with local cascades: Modelling the slow adoption of technology online

Gerardo Iñiguez (CIDE, Mexico/Aalto)

Monday 2016-07-04 11.00 - 12.00

Lecture hall AS3, TUAS building

Global contagion with local cascades: Modelling the slow adoption of technology online

Adoption of innovations, products or online services is commonly interpreted as a spreading process driven to large extent by social influence and conditioned by the needs and capacities of individuals. To model this process one usually introduces behavioural threshold mechanisms, which can give rise to the evolution of global cascades if the system satisfies a set of conditions. However, these models do not address temporal aspects of the emerging cascades, which in real systems may evolve through various pathways ranging from slow to rapid patterns. Here we fill this gap through the analysis and modelling of product adoption in the world’s largest voice over internet service, the social network of Skype. We provide empirical evidence about the heterogeneous distribution of fractional behavioural thresholds, which appears to be independent of the degree of adopting egos. We show that the structure of real-world adoption clusters is radically different from previous theoretical expectations, since vulnerable adoptions—induced by a single adopting neighbour—appear to be important only locally, while spontaneous adopters arriving at a constant rate and the involvement of unconcerned individuals govern the global emergence of social spreading.

Article in Sci. Rep.:
Press release in Aalto:

Journal club on fast route planning algorithms

June 20, 2016 @ 11.00 - 12.00
Lecture room AS3, TUAS
Presenter: Rainer Kujala

Approximate outline of the talk:
1. Graph search basics and terminology
2. (Traditional) Speed-up techniques for static graphs
3. Modern techniques for routing in road (=static networks)
4. Time-dependent road network routing (traffic jams, changing road conditions...)
5. Routing in public transport networks

Based mostly on the article:

Route Planning in Transportation Networks

Authors: Hannah Bast, Daniel Delling, Andrew Goldberg, Matthias Müller-Hannemann, Thomas Pajor, Peter Sanders, Dorothea Wagner, Renato F. Werneck

Abstract: We survey recent advances in algorithms for route planning in transportation networks. For road networks, we show that one can compute driving directions in milliseconds or less even at continental scale. A variety of techniques provide different trade-offs between preprocessing effort, space requirements, and query time. Some algorithms can answer queries in a fraction of a microsecond, while others can deal efficiently with real-time traffic. Journey planning on public transportation systems, although conceptually similar, is a significantly harder problem due to its inherent time-dependent and multicriteria nature. Although exact algorithms are fast enough for interactive queries on metropolitan transit systems, dealing with continent-sized instances requires simplifications or heavy preprocessing. The multimodal route planning problem, which seeks journeys combining schedule-based transportation (buses, trains) with unrestricted modes (walking, driving), is even harder, relying on approximate solutions even for metropolitan inputs.

Download the article here

B. Monastero - Underlined Places. Augmenting a Lobby Floor to Unveil Mobility Patterns and Relations to the Inhabited Space

Beatrice Monastero (Aalto)

Monday 2016-06-13 11.00 - 12.00

Lecture hall AS3, TUAS building

Underlined Places. Augmenting a Lobby Floor to Unveil Mobility Patterns and Relations to the Inhabited Space

I will introduce Underlined Places (UP), an interactive floor projection installed in the lobby of a building. UP displays the mobility patterns of visitors in the building as white traces appearing in real-time on the floor when visitors walk by. Through a 6 week in-field study I investigated how intentional and unintentional interactions with the UP affected awareness of social aspects of the space related to people presence and behaviours. I will present the study and the related results outlining how uses and understanding of the system evolved in time. Three distinct phases of use emerged, characterised by different degrees of appropriation and different attitudes towards the augmented environment of both visitors and employees. Finally I outline how the public reactive visualisation of mobility patterns enhanced social awareness and supported the felt identity of the place (placefulness) and suggest similar urban design interventions.

A. Gabrielli - The Scientific Competitiveness of Nations: a network analysis

Andrea Gabrielli (CNR, Rome)

Monday 2016-06-14 13.00 - 14.00

Lecture hall AS3, TUAS building

The Scientific Competitiveness of Nations: a network analysis

Authors: Andrea Gabrielli, Giulio Cimini, Francesco Sylos Labini, Andrea Zaccaria

We use citation data of scientific articles produced by individual nations in different scientific domains to build a bipartite country - scientific domains network to determine the structure and efficiency of national research systems [1]. We characterize the scientific fitness of each nation—that is, the competitiveness of its research system—and the complexity of each scientific domain by means of a non-linear iterative algorithm [2] able to assess quantitatively the advantage of scientific diversification. We find that technological leading nations, beyond having the largest production of scientific papers and the largest number of citations, do not specialize in a few scientific domains. Rather, they diversify as much as possible their research system. On the other side, less developed nations are competitive only in scientific domains where also many other nations are present. Diversification thus represents the key element that correlates with scientific and technological competitiveness. A remarkable implication of this structure of the scientific competition is that the scientific domains playing the role of “markers” of national scientific competitiveness are those not necessarily of high technological requirements, but rather addressing the most ‘‘sophisticated’’ needs of the society. We complement this analysis with a correlation study between the scientific impact of a nation with a normalized measure of RD funds and the level of internationalization [3].

[1] G. Cimini, A. Gabrielli, F. Sylos Labini (2014), PLoS ONE 9(12), e113470.

[2] A. Tacchella et al. (2013), Sci. Rep. 2, 723.

[3] G. Cimini, A. Zaccaria, A. Gabrielli (2016), J. of Informetrics 10, 200.

Abstract for the talk.

S. Ranganathan - Dynamics of Correlated Investors Network

Sindhuja Ranganathan (TUT)

Monday 2016-06-06 11.00 - 12.00

Lecture hall AS3, TUAS building

Dynamics of Correlated Investors Network

Complex networks methodology has been used in the past to analyse stock markets by analysing correlation networks constructed between stocks. Networks constructed between investors have received considerably less attention in the literature perhaps because of the lack of suitable data sources. In this talk I will outline how to construct and analyse investor networks from large-scale data on Finnish stock markets. The main objective of this research is to analyse the change of the investor behaviour around financial crises, and I will go through some of my preliminary results on longitudinal analysis of the investor networks.

K. Bhattacharya - Quantifying social compensation under risks of relationship weakening

Kunal Bhattacharya (Aalto)

Monday 2016-05-23 11.00 - 12.00

Lecture hall AS3, TUAS building

Quantifying social compensation under risks of relationship weakening

Social networks require active relationship maintenance if they are to be kept at a constant level of emotional closeness. For group-living animals, including humans, failure to interact leads inexorably to a decline in relationship quality, and a consequent loss of the benefits that derive from individual relationships. As a result, many social species compensate for weakened relationships by investing more heavily in them. Here we study how humans behave in similar situations, using data from mobile call detail records from a European country. For the less frequent contacts between pairs of communicating individuals we observe a logarithmic dependence of the duration of the succeeding call on the time gap with the previous call. We find that such behaviour is likely when the individuals in these dyadic pairs have the same gender and are in the same age bracket as well as being geographically distant. Our results indicate that these pairs deliberately invest more time in communication so as to reinforce their social bonding and prevent their relationships decaying when these are threatened by lack of interaction.

M. Nelimarkka - Asking societal questions – a take from political scientist

Matti Nelimarkka (HIIT)

Monday 2016-05-30 12.00 - 13.00

Lecture room AS3, TUAS

Asking societal questions – a take from political scientist

It seems that computational social science is done to demonstrate skills in computation. What kind of collaboration might be needed to improve the situation and what are the challenges in such collaborations? I will reflect the challenges from Digivaalit 2015 and other computational social science projects done together with University of Helsinki, Faculty of Social Science.

C. Purcell - Necessary conditions for efficient role colouring

Christopher Purcell (Aalto University)

Monday 2016-05-16 11.00 - 12.00

Lecture room AS3, TUAS building

Necessary conditions for efficient role colouring

A role colouring of a graph is an assignment of colours to its vertices such that two vertices having the same colour have the same set of colours in their respective neighbourhoods. In a network, nodes with similar roles may have similar neighbourhoods in one way or another. A role colouring is an attempt to formalise an aspect of this phenomenon. This talk will focus on the computational complexity of finding such a colouring with a given number of colours. In particular, we have a sequence of infinitely narrowing classes of graphs for which the problem remains NP-hard, and therefore a necessary condition for a polynomial-time solution in a restricted graph class. Joint work with Puck Rombach.

This seminar is organized jointly with the Large Structures seminar.

Opinions and hierarchies: an agent based model

Jan Snellman (Aalto University)

Monday 2016-05-09 11.00 - 12.00

Lecture hall AS3

Opinions and hierarchies: an agent based model

Will to compete, striving for better status and envy are well known human characteristics. Human societies, on the other hand, are hierarchical by character. The purpose of this research project is to study, whether one can mimic the behaviour of both human societies and individuals in agent based models if one assumes that the main motivation of humans is to maximize their status relative to other humans.

O. Korhonen - How consistent are ROIs as nodes of functional brain networks?

Onerva Korhonen (Aalto University)

Monday 2016-05-02 11.00 - 12.00

Seminar room 1021-1022, TUAS building

How consistent are ROIs as nodes of functional brain networks?

The functional network approach, where fMRI BOLD time series are mapped to complex networks depicting functional relationships between brain areas, has opened new insights into the structure and function of the human brain. In this approach, choosing what the network nodes represent is of crucial importance. One option is to consider individual fMRI voxels as nodes. This results in a large number of nodes and may make network analysis challenging, not only in terms of computational cost but also in terms of interpretation of results. A common alternative is to build the nodes of functional brain networks from pre-defined clusters of anatomically close voxels, known as Regions of Interest (ROIs). This approach assumes that voxels within ROIs are functionally similar and have reasonably similar dynamics, which may not always be true. Because these two approaches are known to result in dissimilar network structures, we argue that it is crucial to understand what happens to network connectivity when moving from the voxel level to the ROI level. We study the consistency of ROIs, defined as the mean Pearson correlation coefficient between the time series of voxels within a ROI, and show that this consistency varies widely in resting-state experimental data. This indicates that the underlying assumption of similar voxel dynamics within each ROI may not generally hold. We then show that this variation is reflected in network properties: consistency correlates with node connectivity and link weight in ROI-level networks.