The dimensionality of biological data is often very high. Feature selection can be used to tackle the problem of high dimensionality. However, majority of the work in feature selection consists of supervised feature selection methods which require class labels. The problem further escalates when the data is time-series gene expression measurements that measure the effect of external stimuli on biological system. In this paper we propose an unsupervised method for gene selection from time-series gene expression data founded on statistical significance testing and swap randomization. We perform experiments with a publicly available mouse gene expression dataset and also a human gene expression dataset describing the exposure to asbestos. The results in both datasets show a considerable decrease in number of genes.

Self-directed lymphocytes may evade clonal deletion at ontogenesis but still remain harmless due to a mechanism called clonal anergy. For B-lymphocytes, two major explanations for anergy developed over the last decades: according to Varela theory, anergy stems from a proper orchestration of the whole B-repertoire, such that self-reactive clones, due to intensive feed-back from other clones, display strong inertia when mounting a response. Conversely, according to the model of cognate response, self-reacting cells are not stimulated by helper lymphocytes and the absence of such signaling yields anergy. Through statistical mechanics we show that helpers do not prompt activation of a sub-group of B-cells: remarkably, the latter are just those broadly interacting in the idiotypic network. Hence Varela theory can finally be reabsorbed into the prevailing framework of the cognate response model. Further, we show how the B-repertoire architecture may emerge, where highly connected clones are self-directed as a natural consequence of ontogenetic learning. 

We study the performance of stochastic local search algorithms for random instances of the K-satisfiability (K-SAT) problem. We present a stochastic local search algorithm, ChainSAT, which moves in the energy landscape of a problem instance by never going upwards in energy. ChainSAT is a focused algorithm in the sense that it focuses on variables occurring in unsatisfied clauses. We show by extensive numerical investigations that ChainSAT and other focused algorithms solve large K-SAT instances almost surely in linear time, up to high clause-to-variable ratios a; for example, for K = 4 we observe linear-time performance well beyond the recently postulated clustering and condensation transitions in the solution space. The performance of ChainSAT is a surprise given that by design the algorithm gets trapped into the first local energy minimum it encounters, yet no such minima are encountered. We also study the geometry of the solution space as accessed by stochastic local search algorithms.

Kindlin proteins represent a novel family of evolutionarily conserved FERM domain containing proteins (FDCPs) and are members of B4.1 superfamily. Kindlins consist of three conserved protein homologs in vertebrates: Kindlin-1, Kindlin-2 and Kindlin-3. All three homologs are associated with focal adhesions and are involved in Integrin activation. FERM domain of each Kindlin is bipartite and plays a key role in Integrin activation. A single ancestral Kindlin protein can be traced back to earliest metazoans, e.g., to Parazoa. This protein underwent multiple rounds of duplication in vertebrates, leading to the present Kindlin family. In this study, we trace phylogenetic and evolutionary history of Kindlin FERM domain with respect to FERM domain of other FDCPs. We show that FERM domain in Kindlin homologs is conserved among Kindlins but amount of conservation is less in comparison with FERM domain of other members in B4.1 superfamily. Furthermore, insertion of Pleckstrin Homology like domain in Kindlin FERM domain has important evolutionary and functional consequences. Important residues in Kindlins are traced and ranked according to their evolutionary significance. The structural and functional significance of high ranked residues is highlighted and validated by their known involvement in Kindlin associated diseases. In light of these findings, we hypothesize that FERM domain originated from a proto-Talin protein in unicellular or proto-multicellular organism and advent of multi-cellularity was accompanied by burst of FDCPs, which supported multi-cellularity functions required for complex organisms. This study helps in developing a better understanding of evolutionary history of FERM domain of FDCPs and the role of FERM domain in metazoan evolution.

Background: Clustering sequences into families has long been an important step in characterization of genes and proteins. There are many algorithms developed for this purpose, most of which are based on either direct similarity between gene pairs or some sort of network structure, where weights on edges of constructed graphs are based on similarity. However, conserved synteny is an important signal that can help distinguish homology and it has not been utilized to its fullest potential. Results: Here, we present GenFamClust, a pipeline that combines the network properties of sequence similarity and synteny to assess homology relationship and merge known homologs into groups of gene families. GenFamClust identifies homologs in a more informed and accurate manner as compared to similarity based approaches. We tested our method against the Neighborhood Correlation method on two diverse datasets consisting of fully sequenced genomes of eukaryotes and synthetic data. Conclusions: The results obtained from both datasets confirm that synteny helps determine homology and GenFamClust improves on Neighborhood Correlation method. The accuracy as well as the definition of synteny scores is the most valuable contribution of GenFamClust.

This work presents two mechanisms for processing fingerprint images; shape-adapted smoothing based on second moment descriptors and automatic scale selection based on normalized derivatives. The shape adaptation procedure adapts the smoothing operation to the local ridge structures, which allows interrupted ridges to be joined without destroying essential singularities such as branching points and enforces continuity of their directional fields. The Scale selection procedure estimates local ridge width and adapts the amount of smoothing to the local amount of noise. In addition, a ridgeness measure is defined, which reflects how well the local image structure agrees with a qualitative ridge model, and is used for spreading the results of shape adaptation into noisy areas. The combined approach makes it possible to resolve fine scale structures in clear areas while reducing the risk of enhancing noise in blurred or fragmented areas. The result is a reliable and adaptively detailed estimate of the ridge orientation field and ridge width, as well as a Smoothed grey-level version of the input image. We propose that these general techniques should be of interest to developers of automatic fingerprint identification systems as well as in other applications of processing related types of imagery.

This work presents a novel technique for preprocessing fingerprint images. The method is based on the measurements of second moment descriptors and shape adaptation of scale-space operators with automatic scale selection (Lindeberg 1994). This procedure, which has been successfully used in the context of shape-from-texture and shape from disparity gradients, has several advantages when applied to fingerprint image enhancement, as observed by (Weickert 1995). For example, it is capable of joining interrupted ridges, and enforces continuity of their directional fields.

In this work, these abovementioned general ideas are applied and extended in the following ways: Two methods for estimating local ridge width are explored and tuned to the problem of fingerprint enhancement. A ridgeness measure is defined, which reflects how well the local image structure agrees with a qualitative ridge model. This information is used for guiding a scale-selection mechanism, and for spreading the results of shape adaptation into noisy areas.

The combined approach makes it possible to resolve fine scale structures in clear areas while reducing the risk of enhancing noise in blurred or fragmented areas. To a large extent, the scheme has the desirable property of joining interrupted lines without destroying essential singularities such as branching points. Thus, the result is a reliable and adaptively detailed estimate of the ridge orientation field and ridge width, as well as a smoothed grey-level version of the input image.

A detailed experimental evaluation is presented, including a comparison with other techniques. We propose that the techniques presented provide mechanisms of interest to developers of automatic fingerprint identification systems.

In phylogenetic foot-printing, putative regulatory elements are found in upstream regions of orthologous genes by searching for common motifs. Motifs in different upstream sequences are subject to mutations along the edges of the corresponding phylogenetic tree, consequently taking advantage of the tree in the motif search is an appealing idea. We describe the Motif Yggdrasil sampler; the first Gibbs sampler based on a general tree that uses unaligned sequences. Previous tree-based Gibbs samplers have assumed a star-shaped tree or partially aligned upstream regions. We give a probabilistic model (MY model) describing upstream sequences with regulatory elements and build a Gibbs sampler with respect to this model. The model allows toggling, i.e., the restriction of a position to a subset of nucleotides, but does not require aligned sequences nor edge lengths, which may be difficult to come by. We apply the collapsing technique to eliminate the need to sample nuisance parameters, and give a derivation of the predictive update formula. We show that the MY model improves the modeling of difficult motif instances and that the use of the tree achieves a substantial increase in nucleotide level correlation coefficient both for synthetic data and 37 bacterial lexA genes. We investigate the sensitivity to errors in the tree and show that using random trees MY sampler still has a performance similar to the original version.

The discriminatory power of the noncoding control region (CR) of domestic dog mitochondrial DNA alone is relatively low. The extent to which the discriminatory power could be increased by analyzing additional highly variable coding regions of the mitochondrial genome (mtGenome) was therefore investigated. Genetic variability across the mtGenome was evaluated by phylogenetic analysis, and the three most variable similar to 1kb coding regions identified. We then sampled 100 Swedish dogs to represent breeds in accordance with their frequency in the Swedish population. A previously published dataset of 59 dog mtGenomes collected in the United States was also analyzed. Inclusion of the three coding regions increased the exclusion capacity considerably for the Swedish sample, from 0.920 for the CR alone to 0.964 for all four regions. The number of mtDNA types among all 159 dogs increased from 41 to 72, the four most frequent CR haplotypes being resolved into 22 different haplotypes.

This thesis gives an introduction to a currently active area in the cross-section between theoretical computer science and theoretical physics. In the last ten years it has been suggested that critical behaviour, usually seen in models from condensed matter physics, may be responsible for the intractability of NP complete computation problems. This would suggest a very deep connection between the two fields on the most fundamental level. How deep this connection really is is subject to ongoing research as well as the topic of this thesis. Some of the conjectrues from the physics community regarding computational hardness in certain problem classes has turned out to be wrong or misinterpreted but the gained interest in both fields has promising potiential in moving the research frontier forward.

The material presented in this thesis is the result of nearly two years work in trying to clearify how the results from physics can be interpreted in the language of actuall computation problems.

We study numerically the solution space structure of random 3-SAT problems close to the SAT/UNSAT transition. This is done by considering chains of satisfiability problems, where clauses are added sequentially to a problem instance. Using the overlap measure of similarity between different solutions found on the same problem instance, we examine geometrical changes as a function of α. In each chain, the overlap distribution is first smooth, but then develops a tiered structure, indicating that the solutions are found in well separated clusters. On chains of not too large instances, all remaining solutions are eventually observed to be found in only one small cluster before vanishing. This condensation transition point is estimated by finite size scaling to be α_c = 4.26 with an apparent critical exponent of about 1.7. The average overlap value is also observed to increase with α up to the transition, indicating a reduction in solutions space size, in accordance with theoretical predictions. The solutions are generated by a local heuristic, ASAT, and compared to those found by the Survey Propagation algorithm up to α_c.

Motivation: Comparative genomics in general and orthology analysis in particular are becoming increasingly important parts of gene function prediction. Previously, orthology analysis and reconciliation has been performed only with respect to the parsimony model. This discards many plausible solutions and sometimes precludes finding the correct one. In many other areas in bioinformatics probabilistic models have proven to be both more realistic and powerful than parsimony models. For instance, they allow for assessing solution reliability and consideration of alternative solutions in a uniform way. There is also an added benefit in making model assumptions explicit and therefore making model comparisons possible. For orthology analysis, uncertainty has recently been addressed using parsimonious reconciliation combined with bootstrap techniques. However, until now no probabilistic methods have been available.

Results: We introduce a probabilistic gene evolution model based on a birth-death process in which a gene tree evolves ‘inside’ a species tree. Based on this model, we develop a tool with the capacity to perform practical orthology analysis, based on Fitch’s original definition, and more generally for reconciling pairs of gene and species trees. Our gene evolution model is biologically sound (Nei et al., 1997) and intuitively attractive. We develop a Bayesian analysis based on MCMC which facilitates approximation of an a posteriori distribution for reconciliations. That is, we can find the most probable reconciliations and estimate the probability of any reconciliation, given the observed gene tree. This also gives a way to estimate the probability that a pair of genes are orthologs. The main algorithmic contribution presented here consists of an algorithm for computing the likelihood of a given reconciliation. To the best of our knowledge, this is the first successful introduction of this type of probabilistic methods, which flourish in phylogeny analysis, into reconciliation and orthology analysis. The MCMC algorithm has been implemented and, although not yet being in its final form, tests show that it performs very well on synthetic as well as biological data. Using standard correspondences, our results carry over to allele trees as well as biogeography.

Phylogeny is both a fundamental tool in biology and a rich source of fascinating modeling and algorithmic problems. Today's wealth of sequenced genomes makes it increasingly important to understand evolutionary events such as duplications, losses, transpositions, inversions, lateral transfers, and domain shuffling. We focus on the gene duplication event, that constitutes a major force in the creation of genes with new function [Ohno 1970; Lynch and Force 2000] and, thereby also, of biodiversity. We introduce the probabilistic gene evolution model, which describes how a gene tree evolves within a given species tree with respect to speciation, gene duplication, and gene loss. The actual relation between gene tree and species tree is captured by a reconciliation, a concept which we generalize for more expressiveness. The model is a canonical generalization of the classical linear birth-death process, obtained by replacing the interval where the process takes place by a tree. For the gene evolution model, we derive efficient algorithms for some associated probability distributions: the probability of a reconciled tree, the probability of a gene tree, the maximum probability reconciliation, the posterior probability of a reconciliation, and sampling reconciliations with respect to the posterior probability. These algorithms provides the basis for several applications, including species tree construction, reconciliation analysis, orthology analysis, biogeography, and host-parasite co-evolution.

This thesis deals with machine learning techniques for the extraction of structure and the analysis of the vertebrate olfactory pathway based on related methods. Some of its main contributions are summarized below.

We have performed a systematic investigation for classification in biomedical images with the goal of recognizing a material in these images by its texture. This investigation included (i) different measures for evaluating the importance of image descriptors (features), (ii) methods to select a feature set based on these evaluations, and (iii) classification algorithms. Image features were evaluated according to their estimated relevance for the classification task and their redundancy with other features. For this purpose, we proposed a framework for relevance and redundancy measures and, within this framework, we proposed two new measures. These were the value difference metric and the fit criterion. Both measures performed well in comparison with other previously used ones for evaluating features. We also proposed a Hopfield network as a method for feature selection, which in experiments gave one of the best results relative to other previously used approaches.

We proposed a genetic algorithm for clustering and tested it on several realworld datasets. This genetic algorithm was novel in several ways, including (i) the use of intra-cluster distance as additional optimization criterion, (ii) an annealing procedure, and (iii) adaptation of mutation rates. As opposed to many conventional clustering algorithms, our optimization framework allowed us to use different cluster validation measures including those which do not rely on cluster centroids. We demonstrated the use of the clustering algorithm experimentally with several cluster validity measures as optimization criteria. We compared the performance of our clustering algorithm to that of the often-used fuzzy c-means algorithm on several standard machine learning datasets from the University of California/Urvine (UCI) and obtained good results.

The organization of representations in the brain has been observed at several stages of processing to spatially decompose input from the environment into features that are somehow relevant from a behavioral or perceptual standpoint. For the perception of smells, the analysis of such an organization, however, is not as straightforward because of the missing metric. Some studies report spatial clusters for several combinations of physico-chemical properties in the olfactory bulb at the level of the glomeruli. We performed a systematic study of representations based on a dataset of activity-related images comprising more than 350 odorants and covering the whole spatial array of the first synaptic level in the olfactory system. We found clustered representations for several physico-chemical properties. We compared the relevance of these properties to activations and estimated the size of the coding zones. The results confirmed and extended previous studies on olfactory coding for physico-chemical properties. Particularly of interest was the spatial progression by carbon chain that we found. We discussed our estimates of relevance and coding size in the context of processing strategies. We think that the results obtained in this study could guide the search into olfactory coding primitives and the understanding of the stimulus space.

In a second study on representations in the olfactory bulb, we grouped odorants together by perceptual categories, such as floral and fruity. By the application of the same statistical methods as in the previous study, we found clustered zones for these categories. Furthermore, we found that distances between spatial representations were related to perceptual differences in humans as reported in the literature. This was possibly the first time that such an analysis had been done. Apart from pointing towards a spatial decomposition by perceptual dimensions, results indicate that distance relationships between representations could be perceptually meaningful.

In a third study, we modeled axon convergence from olfactory receptor neurons to the olfactory bulb. Sensory neurons were stimulated by a set of biologically-relevant odors, which were described by a set of physico-chemical properties that covaried with the neural and glomerular population activity in the olfactory bulb. Convergence was mediated by the covariance between olfactory neurons. In our model, we could replicate the formation of glomeruli and concentration coding as reported in the literature, and further, we found that the spatial relationships between representational zones resulting from our model correlated with reported perceptual differences between odor categories. This shows that natural statistics, including similarity of physico-chemical structure of odorants, can give rise to an ordered arrangement of representations at the olfactory bulb level where the distances between representations are perceptually relevant.

The main problem with sensory processing is the difficulty in relating sensory input to physiological responses and perception. This is especially problematic at higher levels of processing, where complex cues elicit highly specific responses. In olfaction, this relationship is particularly obfuscated by the difficulty of characterizing stimulus statistics and perception. The core questions in olfaction are hence the so-called stimulus problem, which refers to the understanding of the stimulus, and the structure–activity and structure–odor relationships, which refer to the molecular basis of smell. It is widely accepted that the recognition of odorants by receptors is governed by the detection of physico-chemical properties and that the physical space is highly complex. Not surprisingly, ideas differ about how odor stimuli should be classified and about the very nature of information that the brain extracts from odors. Even though there are many measures for smell, there is none that accurately describes all aspects of it. Here, we summarize recent developments in the understanding of olfaction. We argue that an approach to olfactory function where information processing is emphasized could contribute to a high degree to our understanding of smell as a perceptual phenomenon emerging from neural computations. Further, we argue that combined analysis of the stimulus, biology, physiology, and behavior and perception can provide new insights into olfactory function. We hope that the reader can use this review as a competent guide and overview of research activities in olfactory physiology, psychophysics, computation, and psychology. We propose avenues for research, particularly in the systematic characterization of receptive fields and of perception.

The organization of representations in the brain has been observed to locally reflect subspaces of inputs that are relevant to behavioral or perceptual feature combinations, such as in areas receptive to lower and higher-order features in the visual system. The early olfactory system developed highly plastic mechanisms and convergent evidence indicates that projections from primary neurons converge onto the glomerular level of the olfactory bulb (OB) to form a code composed of continuous spatial zones that are differentially active for particular physico?-chemical feature combinations, some of which are known to trigger behavioral responses. In a model study of the early human olfactory system, we derive a glomerular organization based on a set of real-world,biologically-relevant stimuli, a distribution of receptors that respond each to a set of odorants of similar ranges of molecular properties, and a mechanism of axon guidance based on activity. Apart from demonstrating activity-dependent glomeruli formation and reproducing the relationship of glomerular recruitment with concentration, it is shown that glomerular responses reflect similarities of human odor category perceptions and that further, a spatial code provides a better correlation than a distributed population code. These results are consistent with evidence of functional compartmentalization in the OB and could suggest a function for the bulb in encoding of perceptual dimensions.

In university programming classes often hundreds of students participate having to solveeach hundreds of programming assignments a situation which puts instructors to the difficult task of validating hundreds of programming assignments. We present a framework thatcan help instructors and students in organization and validation of program code. Our “System for Automated Assistance in Correction of Programming Exercises“ (short: SAC) is aweb-platform for test-driven development and automated validation. The web-platform isbased on Java Server Pages technology with tomcat as servlet container, and allows teachersto specify and define program exercises and students to upload their solutions. Students can get immediate feedback on the validity of their code and both instructors and students cansee statistics about each programming assignment. We explain our platform and proposehow the automatic validation can be extended. 

This paper aims to introduce Distributed Systems as a field where the ideas and methods of physics can potentially be applied, and to provide entry points to a wide literature. The contributions of Leslie Lamport, inspired by Relativity Theory, and of Edsger Dijkstra, which has the flavor of a growth process, are discussed at some length. The intent of the author is primarily to stimulate interest in the statistical physics community, and the discussions are therefore framed in a non-technical language; the author apologizes in advance to readers from the computer science side for the unavoidable impreciseness and ambiguities.

We present an alternate method to close the master equation representing the continuous time dynamics of interacting Ising spins. The method makes use of the theory of random point processes to derive a master equation for local conditional probabilities. We analytically test our solution studying two known cases, the dynamics of the mean-field ferromagnet and the dynamics of the one-dimensional Ising system. We present numerical results comparing our predictions with Monte Carlo simulations in three different models on random graphs with finite connectivity: the Ising ferromagnet, the random field Ising model, and the Viana-Bray spin-glass model.

The change of the von Neumann entropy of a set of harmonic oscillators initially in thermal equilibrium and interacting linearly with an externally driven quantum system is computed by adapting the Feynman-Vernon influence functional formalism. This quantum entropy production has the form of the expectation value of three functionals of the forward and backward paths describing the system history in the Feynman-Vernon theory. In the classical limit of Kramers-Langevin dynamics (Caldeira-Leggett model) these functionals combine to three terms, where the first is the entropy production functional of stochastic thermodynamics, the classical work done by the system on the environment in units of k(B)T, and the second and the third other functionals which have no analogue in stochastic thermodynamics.

We show that a method based on logistic regression, using all the data, solves the inverse Ising problem far better than mean-field calculations relying only on sample pairwise correlation functions, while still computationally feasible for hundreds of nodes. The largest improvement in reconstruction occurs for strong interactions. Using two examples, a diluted Sherrington-Kirkpatrick model and a two-dimensional lattice, we also show that interaction topologies can be recovered from few samples with good accuracy and that the use of l(1) regularization is beneficial in this process, pushing inference abilities further into low-temperature regimes.

Transcription regulation is largely governed by the profile and the dynamics of transcription factors' binding to DNA. Stochastic effects are intrinsic to this dynamics, and the binding to functional sites must be controlled with a certain specificity for living organisms to be able to elicit specific cellular responses. Specificity stems here from the interplay between binding affinity and cellular abundance of transcription factor proteins, and the binding of such proteins to DNA is thus controlled by their chemical potential. We combine large-scale protein abundance data in the budding yeast with binding affinities for all transcription factors with known DNA binding site sequences to assess the behavior of their chemical potentials in an exponential growth phase. A sizable fraction of transcription factors is apparently bound non-specifically to DNA, and the observed abundances are marginally sufficient to ensure high occupations of the functional sites. We argue that a biological cause of this feature is related to its noise-filtering consequences: abundances below physiological levels do not yield significant binding of functional targets and mis-expressions of regulated genes may thus be tamed.

We establish a refined version of the Second Law of Thermodynamics for Langevin stochastic processes describing mesoscopic systems driven by conservative or non-conservative forces and interacting with thermal noise. The refinement is based on the Monge-Kantorovich optimal mass transport and becomes relevant for processes far from quasi-stationary regime. General discussion is illustrated by numerical analysis of the optimal memory erasure protocol for a model for micron-size particle manipulated by optical tweezers.

We study stationary states in a diluted asymmetric (kinetic) Ising model. We apply the recently introduced dynamic cavity method to compute magnetizations of these stationary states. Depending on the update rule, different versions of the dynamic cavity method apply. We here study synchronous updates and random sequential updates, and compare local properties computed by the dynamic cavity method to numerical simulations. Using both types of updates, the dynamic cavity method is highly accurate at high enough temperatures. At low enough temperatures, for sequential updates the dynamic cavity method tends to a fixed point, but this does not agree with numerical simulations, while for parallel updates, the dynamic cavity method may display oscillatory behavior. When it converges and is accurate, the dynamic cavity method offers a huge speed-up compared to Monte Carlo, particularly for large systems.

We compare dynamic mean-field and dynamic cavity methods to describe the stationary states of dilute kinetic Ising models. We compute dynamic mean-field theory by expanding in interaction strength to third order, and we compare to the exact dynamic mean-field theory for fully asymmetric networks. We show that in diluted networks, the dynamic cavity method generally predicts magnetizations of individual spins better than both first-order ("naive") and second-order ("TAP") dynamic mean-field theory.

We study the dynamic cavity method for dilute kinetic Ising models with synchronous update rules. For he parallel update rule we find for fully asymmetric models that the dynamic cavity equations reduce to a Markovian dynamics of the (time-dependent) marginal probabilities. For the random sequential update rule, also an instantiation of a synchronous update rule, we find on the other hand that the dynamic cavity equations do not reduce to a Markovian dynamics, unless an additional assumption of time factorization is introduced. For symmetric models we show that a fixed point of ordinary Belief propagation is also a fixed point of the dynamic cavity equations in the time factorized approximation. For clarity, the conclusions of the paper are formulated as three lemmas.

We study the problem of optimizing released heat or dissipated work in stochastic thermodynamics. In the overdamped limit these functionals have singular solutions, previously interpreted as protocol jumps. We show that a regularization, penalizing a properly defined acceleration, changes the jumps into boundary layers of finite width. We show that in the limit of vanishing boundary layer width no heat is dissipated in the boundary layer, while work can be done. We further give an alternative interpretation of the fact that the optimal protocols in the overdamped limit are given by optimal deterministic transport (Burgers equation).

Thermodynamics of small systems has become an important field of statistical physics. Such systems are driven out of equilibrium by a control, and the question is naturally posed how such a control can be optimized. We show that optimization problems in small system thermodynamics are solved by (deterministic) optimal transport, for which very efficient numerical methods have been developed, and of which there are applications in cosmology, fluid mechanics, logistics, and many other fields. We show, in particular, that minimizing expected heat released or work done during a nonequilibrium transition in finite time is solved by the Burgers equation and mass transport by the Burgers velocity field. Our contribution hence considerably extends the range of solvable optimization problems in small system thermodynamics.

We investigate the optimal strategy over a finite time horizon for a portfolio of stock and bond and a derivative in an multiplicative Markovian market model with transaction costs (friction). The optimization problem is solved by a Hamilton-Bellman-Jacobi equation, which by the verification theorem has well-behaved solutions if certain conditions on a potential are satisfied. In the case at hand, these conditions simply imply arbitrage-free ("Black-Scholes") pricing of the derivative. While pricing is hence not changed by friction allow a portfolio to fluctuate around a delta hedge. In the limit of weak friction, we determine the optimal control to essentially be of two parts: a strong control, which tries to bring the stock-and-derivative portfolio towards a Black-Scholes delta hedge; and a weak control, which moves the portfolio by adding or subtracting a Black-Scholes hedge. For simplicity we assume growth-optimal investment criteria and quadratic friction.

We propose an alternative notion of time reversal in open quantum systems as represented by linear quantum operations, and a related generalization of classical entropy production in the environment. This functional is the ratio of the probability to observe a transition between two states under the forward and the time reversed dynamics, and leads, as in the classical case, to fluctuation relations as tautological identities. As in classical dynamics in contact with a heat bath, time reversal is not unique, and we discuss several possibilities. For any bistochastic map its dual map preserves the trace and describes a legitimate dynamics reversed in time, in that case the entropy production in the environment vanishes. For a generic stochastic map we construct a simple quantum operation which can be interpreted as a time reversal. For instance, the decaying channel, which sends the excited state into the ground state with a certain probability, can be reversed into the channel transforming the ground state into the excited state with the same probability.

A system of agents moving along a road in both directions is studied numerically within a cellular-automata formulation. An agent steps to the right with probability q or to the left with 1-q when encountering other agents. Our model is restricted to two agent types, traffic-rule abiders (q=1) and traffic-rule ignorers (q=1/2), and the traffic flow, resulting from the interaction between these two types of agents, which is obtained as a function of density and relative fraction. The risk for jamming at a fixed density, when starting from a disordered situation, is smaller when every agent abides by a traffic rule than when all agents ignore the rule. Nevertheless, the absolute minimum occurs when a small fraction of ignorers are present within a majority of abiders. The characteristic features for the spatial structure of the flow pattern are obtained and discussed.

We study the q-state clock models on heptagonal lattices assigned on a negatively curved surface. We show that the system exhibits three classes of equilibrium phases; in between ordered and disordered phases, an intermediate phase characterized by a diverging susceptibility with no magnetic order is observed at every q >= 2. The persistence of the third phase for all q is in contrast with the disappearance of the counterpart phase in a planar system for small q, which indicates the significance of nonvanishing surface-volume ratio that is peculiar in the heptagonal lattice. Analytic arguments based on Ginzburg-Landau theory and generalized Cayley trees make clear that the two-stage transition in the present system is attributed to an energy gap of spin-wave excitations and strong boundary-spin contributions. We further demonstrate that boundary effects break the mean-field character in the bulk region, which establishes the consistency with results of clock models on boundary-free hyperbolic lattices.

We study low-temperature properties of the XY spin model on a negatively curved surface. Geometric curvature of the surface gives rise to frustration in local spin configuration, which results in the formation of high-energy spin clusters scattered over the system. Asymptotic behavior of the spin-glass susceptibility suggests a zero-temperature glass transition, which is attributed to multiple optimal configurations of spin clusters due to nonzero surface curvature of the system. It implies that a constant ferromagnetic spin interaction on a regular lattice can exhibit glasslike behavior without possessing any disorder if the lattice is put on top of a negatively curved space such as a hyperbolic surface.

A typical Go/No-Go decision is suggested to be implemented in the brain via the activation of the direct or indirect pathway in the basal ganglia. Medium spiny neurons (MSNs) in the striatum, receiving input from cortex and projecting to the direct and indirect pathways express D1 and D2 type dopamine receptors, respectively. Recently, it has become clear that the two types of MSNs markedly differ in their mutual and recurrent connectivities as well as feedforward inhibition from FSIs. Therefore, to understand striatal function in action selection, it is of key importance to identify the role of the distinct connectivities within and between the two types of MSNs on the balance of their activity. Here, we used both a reduced firing rate model and numerical simulations of a spiking network model of the striatum to analyze the dynamic balance of spiking activities in D1 and D2 MSNs. We show that the asymmetric connectivity of the two types of MSNs renders the striatum into a threshold device, indicating the state of cortical input rates and correlations by the relative activity rates of D1 and D2 MSNs. Next, we describe how this striatal threshold can be effectively modulated by the activity of fast spiking interneurons, by the dopamine level, and by the activity of the GPe via pallidostriatal backprojections. We show that multiple mechanisms exist in the basal ganglia for biasing striatal output in favour of either the `Go' or the `No-Go' pathway. This new understanding of striatal network dynamics provides novel insights into the putative role of the striatum in various behavioral deficits in patients with Parkinson's disease, including increased reaction times, L-Dopa-induced dyskinesia, and deep brain stimulation-induced impulsivity.

Following an original idea of F. Guerra, in this notes we analyze the Sherrington-Kirkpatrick model from different perspectives, all sharing the underlying approach which consists in linking the resolution of the sta- tistical mechanics of the model (e.g. solving for the free energy) to well-known partial differential equation (PDE) problems (in suitable spaces). The plan is then to solve the related PDE using techniques involved in their native field and lastly bringing back the solution in the proper statistical mechanics framework. Within this strand, after a streamlined test-case on the Curie-Weiss model to highlight the methods more than the physics behind, we solve the SK both at the replica symmetric and at the 1-RSB level, obtaining the correct expression for the free energy via an analogy to a Fourier equation and for the self-consistencies with an analogy to a Burger equation, whose shock wave develops exactly at critical noise level (triggering the phase transition).

Our approach, beyond acting as a new alternative method (with respect to the standard routes) for tackling the complexity of spin glasses, links symmetries in PDE theory with constraints in statistical mechanics and, as a novel result from the theoretical physics perspective, we obtain a new class of polynomial identities (namely of Aizenman-Contucci type, but merged within the Guerra’s broken replica measures), whose in- terest lies in understanding, via the recent Panchenko breakthroughs, how to force the overlap organization to the ultrametric tree predicted by Parisi. 

Motor synergies have been supposed to simplify motor control [1]-[5]. In order to test it, we exploit the correlations of our hand's joints to discover some underlying simplicity in a complex stream of behavioral actions. Instead of averaging variability out, we take the view that the structure of variability may contain valuable information about the task being performed. Therefore, we asked 7 subjects to interact in 17 daily-life situations and quantified behavior in principled manner using cyber glove technology. We combined Probabilistic Principal Component Analysis (PPCA) with a Bayesian classifier to analyze the data. Our key findings are: 1. we confirmed that hand control is low-dimensional, where 4-5 dimensions were sufficient to explain 80-90% of the variability in the movement data [6]. 2. We established a universally applicable measure of manipulative complexity that allowed us to measure this quantity across vastly different tasks. 3. We discovered that within the first 1000 ms of an action the hand shape already configures itself to vastly different tasks, enabling us to reliable predict the action intention [6]. 4. We suggest how using the statistics of natural finger movements paired with Bayesian latent variable model can be used to infer the movements of missing limbs from existing limbs to control e.g. a prosthetic device. Overall, these predictabilities could be used to build intelligent Neuroprosthetics for lost fingers that implement the task from the movement of the remaining limbs.

References

1. Santello, M., Flanders, M., Soechting, J.F. (1998). Postural hand synergies for tool use. J Neurosci. 18, 10105–10115.
2. Todorov, E., Ghahramani, Z. (2004). Analysis of the synergies underlying complex hand manipulation. Conf Proc. IEEE Eng. Med. Biol. Soc. 6, 4637-4640.
3. Weiss, E., Flanders, M. (2004). Muscular and Postural Synergies of the Human Hand. J. Neurophysiol. 92, 523-535.
4. Tresch, M.C., Cheung, V.C.K., d’Avella, A. (2006). Matrix factorization algorithms for the identification of muscle synergies: evaluation on simulated and experimental data sets. J Neurophysiol. 95, 2199–2212.
5. Faisal, A., Stout D., Apel, J., Bradley, B. (2010). The Manipulative Complexity of Lower Paleolithic Stone Toolmaking. PloS ONE 5(11): e13718.
6. Belić, J.J., Faisal, A.A. (2011). The structured variability of finger motor coordination in daily tasks. BMC Neuroscience, doi:10.1186/1471-2202-12-S1-P102.

The only way we can interact with the world is through movements, and our primary interactions are via the hands, thus any loss of hand function has immediate impact on our quality of life. However, to date it has not been systematically assessed how coordination in the hand's joints affects every day actions. This is important for two fundamental reasons. Firstly, to understand the representations and computations underlying motor control “in-the-wild” situations, and secondly to develop smarter controllers for prosthetic hands that have the same functionality as natural limbs. In this work we exploit the correlation structure of our hand and finger movements in daily-life. The novelty of our idea is that instead of averaging variability out, we take the view that the structure of variability may contain valuable information about the task being performed. We asked seven subjects to interact in 17 daily-life situations, and quantified behavior in a principled manner using CyberGlove body sensor networks that, after accurate calibration, track all major joints of the hand. Our key findings are: (1) We confirmed that hand control in daily-life tasks is very low-dimensional, with four to five dimensions being sufficient to explain 80–90% of the variability in the natural movement data. (2) We established a universally applicable measure of manipulative complexity that allowed us to measure and compare limb movements across tasks. We used Bayesian latent variable models to model the low-dimensional structure of finger joint angles in natural actions. (3) This allowed us to build a naïve classifier that within the first 1000 ms of action initiation (from a flat hand start configuration) predicted which of the 17 actions was going to be executed—enabling us to reliably predict the action intention from very short-time-scale initial data, further revealing the foreseeable nature of hand movements for control of neuroprosthetics and tele operation purposes. (4) Using the Expectation-Maximization algorithm on our latent variable model permitted us to reconstruct with high accuracy (<5–6° MAE) the movement trajectory of missing fingers by simply tracking the remaining fingers. Overall, our results suggest the hypothesis that specific hand actions are orchestrated by the brain in such a way that in the natural tasks of daily-life there is sufficient redundancy and predictability to be directly exploitable for neuroprosthetics.

The basal ganglia (BG) represent subcortical structures considered to be involved in action selection and decision making [1]. Dysfunction of the BG circuitry leads to many motor and cognitive disorders such as Parkinson’s disease (PD), Tourette syndrome, Huntington’s disease, obsessive compulsive disorder and many others. Therefore, we simultaneously recorded local field potentials (LFPs) in primary motor cortex and sensorimotor striatum to study features directly related to healthy versus pathological states such as Parkinson disease and levodopa-induced dyskinesia [2], [3]. The striatum, the input stage of the basal ganglia (BG), is an inhibitory network that contains several distinct cell types and receives massive excitatory inputs from the cortex. Cortex sends direct projections to the striatum, while striatum can affect cortex only indirectly through other BG nuclei and thalamus. Firstly we analyzed spectral characteristics of the obtained signals and observed that during dyskinesia, the most prominent feature was a relative power increase in the high gamma frequency range around 80 Hz, while for PD it was the beta frequency range. Secondly our preliminary results have shown that during both pathological states effective connectivity in terms of Granger causality is bidirectional with an accent on striatal influence on cortex. In the case of dyskinesia we have also found a specifically high increase in effective connectivity at 80 Hz. In order to further understand the 80-Hz phenomenon we have performed cross-frequency analysis across all states and both structures and observed characteristic patterns in the case of dyskinesia in both structures but not in the case of PD and healthy state. We have seen a large relative decrease in the modulation of the amplitude at 80Hz by the phase of low frequency oscillations (up to ~10Hz). It has been suggested that the activity of local neural populations is modulated according to the global neuronal dynamics in the way that populations oscillate and synchronize at lower frequencies and smaller ensembles are active at higher frequencies Our results suggest unexpectedly a lack of coupling between the low frequency activity of a larger population and the synchronized activity of a smaller group of neurons active at 80Hz.

Linkage between behavioral states and neural activity is one of the most important challenges in neuroscience. The network activity patterns in the awake resting state and in the actively behaving state in rodents are not well understood, and a better tool for differentiating these states can provide insights on healthy brain functions and its alteration with disease. Therefore, we simultaneously recorded local field potentials (LFPs) bilaterally in motor cortex and striatum, and measured locomotion from healthy, freely behaving rats. Here we analyze spectral characteristics of the obtained signals and present an algorithm for automatic discrimination of the awake resting and the behavioral states. We used the Support Vector Machine (SVM) classifier and utilized features obtained by applying discrete wavelet transform (DWT) on LFPs, which arose as a solution with high accuracy.

Neuronal avalanches are found in the resting state activity of the mammaliancortex. Here we studied whether and how cortical avalanches are mappedonto the striatal circuitry, the first stage of the basal ganglia. We first demonstrate using organotypic cortex-striatum-substantia nigra cultures from rat that indeed striatal neurons respond to cortical avalanches originating in superficial layers. We simultaneously recorded spontaneous local field potentials (LFPs) in the cortical and striatal tissue using high-density microelectrode arrays. In the cortex, spontaneous neuronal avalanches were characterized by intermittent spatiotemporal activity clusters with a cluster size distribution that followed a power law with exponent 1.5. In the striatum, intermittent spatiotemporal activity was found to correlate with cortical avalanches. However, striatal negative LFP peaks (nLFPs) did not showavalanche signatures, but formed a cluster size distribution that had a much steeper drop-off, i.e., lacked large spatial clusters that are commonly expected for avalanche dynamics. The underlying de-correlation of striatal activity could have its origin in the striatum through local inhibition and/or could result from a particular mapping in the corticostriatal pathway. Here we show, using modeling, that highly convergent corticostriatal projections can map spatially extended cortical activity into spatially restricted striatal regimes.

User of Brain Computer Interface system stays in “idle state” between executions of motor task or imagining one. Movement-based BCIs can operate in synchronous and asynchronous mode and in both cases in order to make the system robust, it is necessary that the system is able to distinguish with certainty idle state from the initiation of the movement. We propose computing method that determines the probability of subject's intention to make movement in comparison to idle state. We therefore asked 4 subjects between 20-30 years of age to perform the task of pressing the taster button by thumb while their EEG recordings were obtained. First, the subjects performed motor task at instants defined with the animation shown on screen and second, subjects performed self-initiated movement. Movement onsets were identified by voltage change when taster sensor was pressed while analysis was based on the Event Related Desynchronisation (ERD). This neurophysiological phenomenon refers to the decrease of the EEG signal power just before the voluntary movement onset (pre-movement state). Features of the extracted signals were determined by applying one of the following methods: Welch's method, Burg's algorithm or wavelet transform. In order to distinguish data in the two states, we performed classification by using Support Vector Machine (SVM) method. Results showed that SVM classifier was able to anticipate up to 78% of the movements executed.

This thesis consists of three parts related to the in silico study of the brain: technologies for large-scale neural simulations, neural algorithms and models and applications in large-scale data analysis in neuroinformatics. All parts rely on the use of supercomputers.

A large-scale neural simulator is developed where techniques are explored for the simulation, analysis and visualization of neural systems on a high biological abstraction level. The performance of the simulator is investigated on some of the largest supercomputers available.

Neural algorithms and models on a high biological abstraction level are presented and simulated. Firstly, an algorithm for structural plasticity is suggested which can set up connectivity and response properties of neural units from the statistics of the incoming sensory data. This can be used to construct biologically inspired hierarchical sensory pathways. Secondly, a model of the mammalian olfactory system is presented where we suggest a mechanism for mixture segmentation based on adaptation in the olfactory cortex. Thirdly, a hierarchical model is presented which uses top-down activity to shape sensory representations and which can encode temporal history in the spatial representations of populations.

Brain-inspired algorithms and methods are applied to two neuroinformatics applications involving large-scale data analysis. In the first application, we present a way to extract resting-state networks from functional magnetic resonance imaging (fMRI) resting-state data where the final extraction step is computationally inexpensive, allowing for rapid exploration of the statistics in large datasets and their visualization on different spatial scales. In the second application, a method to estimate the radioactivity level in arterial plasma from segmented blood vessels from positron emission tomography (PET) images is presented. The method outperforms previously reported methods to a degree where it can partly remove the need for invasive arterial cannulation and continuous sampling of arterial blood during PET imaging.

In conclusion, this thesis provides insights into technologies for the simulation of large-scale neural models on supercomputers, their use to study mechanisms for the formation of neural representations and functions in hierarchical sensory pathways using models on a high biological abstraction level and the use of large-scale, fine-grained data analysis in neuroinformatics applications.

Non-parametric data-driven analysis techniques can be used to study datasets with few assumptions about the data and underlying experiment. Variations of independent component analysis (ICA) have been the methods mostly used on fMRI data, e.g., in finding resting-state networks thought to reflect the connectivity of the brain. Here we present a novel data analysis technique and demonstrate it on resting-state fMRI data. It is a generic method with few underlying assumptions about the data. The results are built from the statistical relations between all input voxels, resulting in a whole-brain analysis on a voxel level. It has good scalability properties and the parallel implementation is capable of handling large datasets and databases. From the mutual information between the activities of the voxels over time, a distance matrix is created for all voxels in the input space. Multidimensional scaling is used to put the voxels in a lower-dimensional space reflecting the dependency relations based on the distance matrix. By performing clustering in this space we can find the strong statistical regularities in the data, which for the resting-state data turns out to be the resting-state networks. The decomposition is performed in the last step of the algorithm and is computationally simple. This opens up for rapid analysis and visualization of the data on different spatial levels, as well as automatically finding a suitable number of decomposition components.

The response properties of sensors such as electronic noses vary in time due to internal or environmental factors. Recalibration is often costly or technically infeasible, which is why algorithms aimed at addressing the sensor drift problem at the data processing level have been developed. These falls in two categories: The pre-processing approaches, such as component correction [1], try to extract the direction and amount of drift in the training data and remove the drift component during operation. Adaptive algorithms, such as the self-organizing map [2], try to counteract the drift during runtime by adjusting the network to the incoming data.

We have previously suggested a modular neural network architecture as a model of cortical layer 4 [3]. Here we show how it quite well can handle the sensor drift problem in chemosensor data. It creates a distributed and redundant code suitable for a noisy and drifting environment. A feature extraction layer governed by competitive learning allows for network adaptation during runtime. In addition, training data can be utilized to create a prediction of the underlying drift to further improve the network performance. Hence, we attempt to combine the two aforementioned methodological categories into one network model.

The capabilities of the proposed network are demonstrated on surrogate data as well as real-world data collected from an electronic nose.