Showing papers by "Helsinki University of Technology published in 2000"

Effects of Age at Entry, Knowledge Intensity, and Imitability on International Growth

Abstract: We employed knowledge-based theory to shed light on international growth in entrepreneurial firms. We found earlier initiation of internationalization and greater knowledge intensity to be associat...

Clustering of the self-organizing map

Abstract: The self-organizing map (SOM) is an excellent tool in exploratory phase of data mining. It projects input space on prototypes of a low-dimensional regular grid that can be effectively utilized to visualize and explore properties of the data. When the number of SOM units is large, to facilitate quantitative analysis of the map and the data, similar units need to be grouped, i.e., clustered. In this paper, different approaches to clustering of the SOM are considered. In particular, the use of hierarchical agglomerative clustering and partitive clustering using K-means are investigated. The two-stage procedure-first using SOM to produce the prototypes that are then clustered in the second stage-is found to perform well when compared with direct clustering of the data and to reduce the computation time.

Self organization of a massive document collection

Abstract: Describes the implementation of a system that is able to organize vast document collections according to textual similarities. It is based on the self-organizing map (SOM) algorithm. As the feature vectors for the documents statistical representations of their vocabularies are used. The main goal in our work has been to scale up the SOM algorithm to be able to deal with large amounts of high-dimensional data. In a practical experiment we mapped 6840568 patent abstracts onto a 1002240-node SOM. As the feature vectors we used 500-dimensional vectors of stochastic figures obtained as random projections of weighted word histograms.

Comparison of model surfaces for cellulose interactions: elevated pH

Abstract: Two different substrates have been used to measure interaction forces between cellulose and between cellulose and glass at normal and high pH. Forces between microspheres of cellulose (r = 20–30 μm) have been measured using the colloidal probe atomic force microscopy technique. Interactions between Langmuir—Blodgett cellulose films on a hydrophobised mica substrate and a glass sphere have been determined with the noninterferometric surface force apparatus. Also, the interaction between two identical Langmuir—Blodgett cellulose films determined with the interferometric surface force apparatus is given for comparison. At low pH (5.5–6) the interaction at large separations in both systems is characterised by a double-layer repulsion with an electrosteric contribution dominating the shorter-range regime. At pH 10, the Langmuir—Blodgett cellulose film swells considerably, which generates a long-range steric repulsion. In many cases several inward steps have been observed in the force—distance curves. We attribute this to a sudden partial collapse of the swollen cellulose film. After initial compression of the steric layer (upon consecutive force runs) the long-range interaction is again dominated by a double-layer force. In contrast, measurements between two cellulose spheres have shown no excessive swelling. Only a limited increase (from about 10 nm to about 20 nm per surface) of the range of the electrosteric repulsion has been found at pH 10. The force at longer distances is in good agreement with the Poisson—Boltzmann theory, with the surface potential increasing with pH as expected.

Independent component approach to the analysis of EEG and MEG recordings

Abstract: Multichannel recordings of the electromagnetic fields emerging from neural currents in the brain generate large amounts of data. Suitable feature extraction methods are, therefore, useful to facilitate the representation and interpretation of the data. Recently developed independent component analysis (ICA) has been shown to be an efficient tool for artifact identification and extraction from electroencephalographic (EEG) and magnetoencephalographic (MEG) recordings. In addition, ICA has been applied to the analysis of brain signals evoked by sensory stimuli. This paper reviews our recent results in this field.

A fast fixed-point algorithm for independent component analysis of complex valued signals.

Abstract: Separation of complex valued signals is a frequently arising problem in signal processing. For example, separation of convolutively mixed source signals involves computations on complex valued signals. In this article, it is assumed that the original, complex valued source signals are mutually statistically independent, and the problem is solved by the independent component analysis (ICA) model. ICA is a statistical method for transforming an observed multidimensional random vector into components that are mutually as independent as possible. In this article, a fast xed-point type algorithm that is capable of separating complex valued, linearly mixed source signals is presented and its computational eciency is shown by simulations. Also, the local consistency of the estimator given by the algorithm is proved.

Emergence of Phase- and Shift-Invariant Features by Decomposition of Natural Images into Independent Feature Subspaces

Abstract: Olshausen and Field (1996) applied the principle of independence maximization by sparse coding to extract features from natural images. This leads to the emergence of oriented linear filters that have simultaneous localization in space and in frequency, thus resembling Gabor functions and simple cell receptive fields. In this article, we show that the same principle of independence maximization can explain the emergence of phase- and shift-invariant features, similar to those found in complex cells. This new kind of emergence is obtained by maximizing the independence between norms of projections on linear subspaces (instead of the independence of simple linear filter outputs). The norms of the projections on such “independent feature subspaces” then indicate the values of invariant features.

Temporal dynamics of cortical representation for action

Abstract: Brain-imaging studies have shown that the human Broca's region and precentral motor cortex are activated both during execution of hand actions and during observation of similar actions performed by other individuals We aimed to clarify the temporal dynamics of this cortical activation by neuromagnetic recordings during execution, on-line imitation, and observation of right-hand reaching movements that ended with a precision pinch of the tip of a manipulandum During execution, the left inferior frontal cortex [Brodmann's area (BA) 44] was activated first (peak approximately 250 ms before the pinching); this activation was followed within 100-200 ms by activation in the left primary motor area (BA4) and 150-250 ms later in the right BA4 During imitation and observation, the sequence was otherwise similar, but it started from the left occipital cortex (BA19) Activation was always strongest during action imitation Only the occipital activation was detected when the subject observed the experimenter reaching his hand without pinching These results suggest that the left BA44 is the orchestrator of the human "mirror neuron system" and is strongly involved in action imitation The mirror system matches action observation and execution and probably contributes to our understanding of actions made by others

Theta oscillations index human hippocampal activation during a working memory task

Abstract: Working memory (WM) is the ability to retain and associate information over brief time intervals. Functional imaging studies demonstrate that WM is mediated by a distributed network including frontal and posterior cortices, hippocampus, and cerebellum. In rodents, the presentation of stimuli in a WM task is followed by a reset of the phase of hippocampal theta. In this paper we report the observation of a similar phenomenon in normal human subjects. Neuromagnetic responses were recorded during presentation of a set of digits and a subsequent probe of the retained items. All stimuli were presented with a fixed temporal pattern. We observed phase reset of ≈7 Hz theta in left hippocampus ≈120 ms after probe stimuli, whereas reset of theta in right hippocampus was visible ≈80 ms prior to these anticipated stimuli. The duration of stimulus-locked theta increased with memory load, with a limiting value of ≈600 ms for 5–7 retained items. We suggest that, as in rats, stimulus-locked theta may index involvement of human hippocampal networks in the cognitive processing of sensory input. The anticipatory phase reset of theta indicates involvement of hippocampus in right hemisphere and cerebellar timing networks. Hippocampal structures are essential for orientation to perturbations in the sensory scene, a function that requires use of a context established by a constellation of stimuli. We suggest that the initiation and maintenance of stimulus-locked hippocampal theta observed here may facilitate processing of potentially salient and/or novel input with respect to a context established by the contents of WM.

Human Cortical Muscle Coherence Is Directly Related to Specific Motor Parameters

Abstract: Cortical oscillations have been the target of many recent investigations, because it has been proposed that they could function to solve the "binding" problem. In the motor cortex, oscillatory activity has been reported at a variety of frequencies between approximately 4 and approximately 60 Hz. Previous research has shown that 15-30 Hz oscillatory activity in the primary motor cortex is coherent or phase locked to activity in contralateral hand and forearm muscles during isometric contractions. However, the function of this oscillatory activity remains unclear. Is it simply an epiphenomenon or is it related to specific motor parameters? In this study, we investigated task-dependent modulation in coherence between motor cortex and hand muscles during precision grip tasks. Twelve right-handed subjects used index finger and thumb to grip two levers that were under robotic control. Each lever was fitted with a sensitive force gauge. Subjects received visual feedback of lever force levels and were instructed to keep them within target boxes throughout each trial. Surface EMGs were recorded from four hand and forearm muscles, and magnetoencephalography (MEG) was recorded using a 306 channel neuromagnetometer. All subjects showed significant levels of coherence (0.086-0.599) between MEG and muscle in the 15-30 Hz range. Coherence was significantly smaller when the task was performed under an isometric condition (levers fixed) compared with a compliant condition in which subjects moved the levers against a spring-like load. Furthermore, there was a positive, significant relationship between the level of coherence and the degree of lever compliance. These results argue in favor of coherence between cortex and muscle being related to specific parameters of hand motor function.

DOS-Resistant Authentication with Client Puzzles

Abstract: Denial of service by server resource exhaustion has become a major security threat in open communications networks. Public-key authentication does not completely protect against the attacks because the authentication protocols often leave ways for an unauthenticated client to consume a server's memory space and computational resources by initiating a large number of protocol runs and inducing the server to perform expensive cryptographic computations. We show how stateless authentication protocols and the client puzzles of Juels and Brainard can be used to prevent such attacks.

Accuracy comparison of various remote sensing data sources in the retrieval of forest stand attributes

Abstract: Recent advances in developing new airborne instruments and space-borne missions and in SAR technology, especially in interferometry and coherence estimation, have roused questions: can such new SAR data be utilized in operational forest inventory? What is the accuracy of different satellite data for forest inventory? This paper verifies the explanatory power and information contents of several remote sensing data sources on the retrieval of stem volume, basal area, and mean height, utilizing the following data: Landsat TM, Spot PAN and XS, ERS-1/2 PRI and SLC (coherence estimation), airborne data from imaging spectrometer AISA, radar-derived forest canopy profiles (obtained with HUTSCAT), and aerial photographs. Ground truth data included three different sets ranging from conventional forest inventory data to intensive field checking where one man-day was spent for assessing one stand. Multivariate and neural network methods were applied in data analysis. The results suggested that (1) radar-derived stand profiles obtained with 100 m spacing was the most accurate data source in this comparison and was of equivalent accuracy with conventional forest inventory for mean height and stem volume estimation, (2) aerial photographs (scale 1 : 20,000) gave comparable results with the imaging spectrometer AISA, (3) the satellite images used for the estimation in the decreasing explanation power were Spot XS, Spot PAN, Landsat TM, ERS SAR coherence, JERS SAR intensity images (PRI); and ERS SAR intensity images (PRI). It appears that optical images still include more information for forest inventory than radar images, (4) from all satellite radar methods, the coherence technique seemed to be superior to other methods.

Mixing Rules with Complex Dielectric Coefficients

Abstract: This article discusses the determination of effective dielectric properties of hetereogeneous materials, in particular media with lossy constituents that have complex permittivity parameters. Several different accepted mixing rules are presented and the effects of the structure and internal geometry of the mixture on the effective permittivity are illustrated. Special attention is paid to phenomena that the mixing process causes in the character of the macroscopic dielectric response of the mixture when the losses of one or several of the components are high or when there is a strong dielectric contrast between the component permittivities.

A computationally efficient multipitch analysis model

Abstract: A computationally efficient model for multipitch and periodicity analysis of complex audio signals is presented. The model essentially divides the signal into two channels, below and above 1000 Hz, computes a "generalized" autocorrelation of the low-channel signal and of the envelope of the high-channel signal, and sums the autocorrelation functions. The summary autocorrelation function (SACF) is further processed to obtain an enhanced SACF (ESACF). The SACF and ESACP representations are used in observing the periodicities of the signal. The model performance is demonstrated to be comparable to those of recent time-domain models that apply a multichannel analysis. In contrast to the multichannel models, the proposed pitch analysis model can be run in real time using typical personal computers. The parameters of the model are experimentally tuned for best multipitch discrimination with typical mixtures of complex tones. The proposed pitch analysis model may be used in complex audio signal processing applications, such as sound source separation, computational auditory scene analysis, and structural representation of audio signals. The performance of the model is demonstrated by pitch analysis examples using sound mixtures which are available for download at http://www.acoustics.hut.fi/-ttolonen/pitchAnalysis/.

Statistical inversion and Monte Carlo sampling methods in electrical impedance tomography

Abstract: This paper discusses the electrical impedance tomography (EIT) problem: electric currents are injected into a body with unknown electromagnetic properties through a set of contact electrodes. The corresponding voltages that are needed to maintain these currents are measured. The objective is to estimate the unknown resistivity, or more generally the impedivity distribution of the body based on this information. The most commonly used method to tackle this problem in practice is to use gradient-based local linearizations. We give a proof for the differentiability of the electrode boundary data with respect to the resistivity distribution and the contact impedances. Due to the ill-posedness of the problem, regularization has to be employed. In this paper, we consider the EIT problem in the framework of Bayesian statistics, where the inverse problem is recast into a form of statistical inference. The problem is to estimate the posterior distribution of the unknown parameters conditioned on measurement data. From the posterior density, various estimates for the resistivity distribution can be calculated as well as a posteriori uncertainties. The search of the maximum a posteriori estimate is typically an optimization problem, while the conditional expectation is computed by integrating the variable with respect to the posterior probability distribution. In practice, especially when the dimension of the parameter space is large, this integration must be done by Monte Carlo methods such as the Markov chain Monte Carlo (MCMC) integration. These methods can also be used for calculation of a posteriori uncertainties for the estimators. In this paper, we concentrate on MCMC integration methods. In particular, we demonstrate by numerical examples the statistical approach when the prior densities are non-differentiable, such as the prior penalizing the total variation or the L1 norm of the resistivity.

Density-functional study of bulk and surface properties of titanium nitride using different exchange-correlation functionals

Abstract: The adsorption and diffusion of hydrogen on the (100) surface of titanium nitride was studied using density-functional theory (DFT) and the generalized gradient approximation (GGA) for the exchange and correlation energy. The adsorption site was found to be on top of the titanium atom with the chemisorption energy of -2.88 eV. The diffusion barrier was determined as 0.73 eV along the path connecting the neighboring titanium atoms. The surface energies and surface relaxations of the three most important surfaces of TiN were studied. The surface energies have the following order: ${S}_{100}l{S}_{110}l{S}_{111}.$ Three different GGA functionals, the Perdew-Wang 1991 (PW91), the Perdew-Burke-Ernzerhof (PBE), and the revised PBE (RPBE) functionals, were tested on crystals, small molecules and TiN surfaces. The RPBE functional when applied to the surface studies of TiN was found to produce slightly lower values of surface energies and of hydrogen adsorption energies than the PW91 functional.

Giant field-induced reversible strain in magnetic shape memory NiMnGa alloy

Abstract: A room temperature extensional strain of 5.1% was observed in martensitic Ni/sub 48/Mn/sub 31/Ga/sub 21/ alloy in the magnetic field of 480 kA/m. The magnitude of field-induced strain decreases with increasing external compressive stress applied in the direction of expansion. The compressive stress of about 3 MPa prevents the development of the substantial field-induced strain. Magnetization curves obtained by VSM exhibit an abrupt magnetization change and a transient hysteresis in the first quadrant. Large reversible field-induced strain and the abrupt magnetization change are due to the rearrangement or redistribution of martensitic twin variants by the applied magnetic field. It was confirmed by optical observation of movement and nucleation of martensitic twin boundaries.

Effective permittivity of mixtures: numerical validation by the FDTD method

Abstract: The present paper reports the results of an extensive numerical analysis of electromagnetic fields in random dielectric materials. The effective permittivity of a two-dimensional (2-D) dielectric mixture is calculated by FDTD simulations of such a sample in a TEM waveguide. Various theoretical bounds are tested in light of the numerical simulations. The results show how the effective permittivity of a mixture with random inclusion positionings is distributed. All possible permittivity values lie between Wiener limits, and according to FDTD simulations the values are almost always between Hashin-Shtrikman limits. Calculated permittivity distribution is also compared with theoretical mixture models. No model seems to be able to predict the simulated behavior over the whole range of volume fraction.

Speaking modifies voice-evoked activity in the human auditory cortex.

Abstract: The voice we most often hear is our own, and proper interaction between speaking and hearing is essential for both acquisition and performance of spoken language. Disturbed audiovocal interactions have been implicated in aphasia, stuttering, and schizophrenic voice hallucinations, but paradigms for a noninvasive assessment of auditory self-monitoring of speaking and its possible dysfunctions are rare. Using magnetoencephalograpy we show here that self-uttered syllables transiently activate the speaker's auditory cortex around 100 ms after voice onset. These phasic responses were delayed by 11 ms in the speech-dominant left hemisphere relative to the right, whereas during listening to a replay of the same utterances the response latencies were symmetric. Moreover, the auditory cortices did not react to rare vowel changes interspersed randomly within a series of repetitively spoken vowels, in contrast to regular change-related responses evoked 100-200 ms after replayed rare vowels. Thus, speaking primes the human auditory cortex at a millisecond time scale, dampening and delaying reactions to self-produced "expected" sounds, more prominently in the speech-dominant hemisphere. Such motor-to-sensory priming of early auditory cortex responses during voicing constitutes one element of speech self-monitoring that could be compromised in central speech disorders.

Cortico-muscular synchronization during isometric muscle contraction in humans as revealed by magnetoencephalography.

Abstract: Magnetoencephalographic (MEG) and electromyographic (EMG) signals were recorded from six subjects during isometric contraction of four different muscles. Cortical sources were located from the MEG signal which was averaged time-locked to the onset of motor unit potentials. A spatial filtering algorithm was used to estimate the source activity. Sources were found in the primary motor cortex (M1) contralateral to the contracted muscle. Significant coherence between rectified EMG and M1 activity was seen in the 20 Hz frequency range in all subjects. Interactions between the motor cortex and spinal motoneuron pool were investigated by separately studying the non-stationary phase and amplitude dynamics of M1 and EMG signals. Delays between M1 and EMG signals, computed from their phase difference, were found to be in agreement with conduction times from the primary motor cortex to the respective muscle. The time-dependent cortico-muscular phase synchronization was found to be correlated with the time course of both M1 and EMG signals. The findings demonstrate that the coupling between the primary motor cortex and motoneuron pool is at least partly due to phase synchronization of 20 Hz oscillations which varies over time. Furthermore, the consistent phase lag between M1 and EMG signals, compatible with conduction time between M1 and the respective muscle with the M1 activity preceding EMG activity, supports the conjecture that the motor cortex drives the motoneuron pool.

Independent component analysis applied to feature extraction from colour and stereo images.

Abstract: Previous work has shown that independent component analysis (ICA) applied to feature extraction from natural image data yields features resembling Gabor functions and simple-cell receptive fields. This article considers the effects of including chromatic and stereo information. The inclusion of colour leads to features divided into separate red/green, blue/yellow, and bright/dark channels. Stereo image data, on the other hand, leads to binocular receptive fields which are tuned to various disparities. The similarities between these results and the observed properties of simple cells in the primary visual cortex are further evidence for the hypothesis that visual cortical neurons perform some type of redundancy reduction, which was one of the original motivations for ICA in the first place. In addition, ICA provides a principled method for feature extraction from colour and stereo images; such features could be used in image processing operations such as denoising and compression, as well as in pattern recognition.

Single word reading in developmental stutterers and fluent speakers

Abstract: Ten fluent speakers and nine developmental stutterers read isolated nouns aloud in a delayed reading paradigm. Cortical activation sequences were mapped with a whole-head magnetoencephalography system. The stutterers were mostly fluent in this task. Although the overt performance was essentially identical in the two groups, the cortical activation patterns showed clear differences, both in the evoked responses, time-locked to word presentation and mouth movement onset, and in task-related suppression of 20-Hz oscillations. Within the first 400 ms after seeing the word, processing in fluent speakers advanced from the left inferior frontal cortex (articulatory programming) to the left lateral central sulcus and dorsal premotor cortex (motor preparation). This sequence was reversed in the stutterers, who showed an early left motor cortex activation followed by a delayed left inferior frontal signal. Stutterers thus appeared to initiate motor programmes before preparation of the articulatory code. During speech production, the right motor/premotor cortex generated consistent evoked activation in fluent speakers but was silent in stutterers. On the other hand, suppression of motor cortical 20-Hz rhythm, reflecting task-related neuronal processing, occurred bilaterally in both groups. Moreover, the suppression was right-hemisphere dominant in stutterers, as opposed to left-hemisphere dominant in fluent speakers. Accordingly, the right frontal cortex of stutterers was highly active during speech production but did not generate synchronous time-locked responses. The speech-related 20-Hz suppression concentrated in the mouth area in fluent speakers, but was evident in both the hand and mouth areas in stutterers. These findings may reflect imprecise functional connectivity within the right frontal cortex and incomplete segregation between the adjacent hand and mouth motor representations in stutterers during speech production. A network including the left inferior frontal cortex and the right motor/premotor cortex, likely to be relevant in merging linguistic and affective prosody with articulation during fluent speech, thus appears to be partly dysfunctional in developmental stutterers.

PicSOM—content-based image retrieval with self-organizing maps

Abstract: We have developed a novel system for content-based image retrieval in large, unannotated databases. The system is called PicSOM, and it is based on tree structured self-organizing maps (TS-SOMs). Given a set of reference images, PicSOM is able to retrieve another set of images which are similar to the given ones. Each TS-SOM is formed with a diAerent image feature representation like color, texture, or shape. A new technique introduced in PicSOM facilitates automatic combination of responses from multiple TS-SOMs and their hierarchical levels. This mechanism adapts to the user’s preferences in selecting which images resemble each other. Thus, the mechanism implements a relevance feedback technique on content-based image retrieval. The image queries are performed through the World Wide Web and the queries are iteratively refined as the system exposes more images to the user. ” 2000 Elsevier Science B.V. All rights reserved.

Components of functional-structural tree models

Abstract: Cet article a pour objet de faire une synthese des approches realisees dans le cadre de la modelisation fonctionnelle et structurale des arbres (FSM). Ces modeles resultent du couplage entre la modelisation du fonctionnement ecophysiologique d'arbres, d'une part, et la modelisation des processus morphologiques, d'autre part. Apres une breve presentation des approches existantes, nous introduisons la notion «d'unite elementaire ideale» (IEU) qui peut etre consideree comme la composante fondamentale des FSM au regard de la souplesse qu'elle confere dans l'articulation des processus. La distribution des metabolites et la croissance sont ensuite abordees comme etant les processus a resoudre de facon prioritaire dans le developpement des FSM, et les differentes approches pouvant etre mises a contribution dans la construction de ces modeles sont discutees. Enfin nous analysons les besoins en programmation des FSM, discutons des avancees necessaires et evaluons leur adequation a la resolution d'objectifs divers.

Web-Hipre: Global Decision Support By Value Tree And AHP Analysis

Abstract: Web-HIPRE is a Java applet for multiple criteria decision analysis. Being located on the WWW, it can be accessed from everywhere in the world. This has opened up a completely new era and dimension in decision support. Web-HIPRE provides a common platform for individual and group decision making. The models can be processed at the same or at different times and the results can be easily shared and combined. There is a possibility to define links to other WWW addresses. These links can refer to any other kind of information such as graphics, sound or video describing the criteria or alternatives. This can improve the quality of decision support dramatically. The most common weighting methods including AHP, SMART, SWING, SMARTER and value functions are supported. Web-HIPRE is located on http://www.hipre.hut.fi/