TY - CONF
AB - We introduce and study a new notion of enhanced chosen-ciphertext security (ECCA) for public-key encryption. Loosely speaking, in the ECCA security experiment, the decryption oracle provided to the adversary is augmented to return not only the output of the decryption algorithm on a queried ciphertext but also of a randomness-recovery algorithm associated to the scheme. Our results mainly concern the case where the randomness-recovery algorithm is efficient. We provide constructions of ECCA-secure encryption from adaptive trapdoor functions as defined by Kiltz et al. (EUROCRYPT 2010), resulting in ECCA encryption from standard number-theoretic assumptions. We then give two applications of ECCA-secure encryption: (1) We use it as a unifying concept in showing equivalence of adaptive trapdoor functions and tag-based adaptive trapdoor functions, resolving an open question of Kiltz et al. (2) We show that ECCA-secure encryption can be used to securely realize an approach to public-key encryption with non-interactive opening (PKENO) originally suggested by Damgård and Thorbek (EUROCRYPT 2007), resulting in new and practical PKENO schemes quite different from those in prior work. Our results demonstrate that ECCA security is of both practical and theoretical interest.
AU - Dachman Soled, Dana
AU - Fuchsbauer, Georg
AU - Mohassel, Payman
AU - O’Neill, Adam
ED - Krawczyk, Hugo
ID - 2045
T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
TI - Enhanced chosen-ciphertext security and applications
VL - 8383
ER -
TY - CONF
AB - We introduce policy-based signatures (PBS), where a signer can only sign messages conforming to some authority-specified policy. The main requirements are unforgeability and privacy, the latter meaning that signatures not reveal the policy. PBS offers value along two fronts: (1) On the practical side, they allow a corporation to control what messages its employees can sign under the corporate key. (2) On the theoretical side, they unify existing work, capturing other forms of signatures as special cases or allowing them to be easily built. Our work focuses on definitions of PBS, proofs that this challenging primitive is realizable for arbitrary policies, efficient constructions for specific policies, and a few representative applications.
AU - Bellare, Mihir
AU - Fuchsbauer, Georg
ED - Krawczyk, Hugo
ID - 2046
T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
TI - Policy-based signatures
VL - 8383
ER -
TY - CONF
AB - Following the publication of an attack on genome-wide association studies (GWAS) data proposed by Homer et al., considerable attention has been given to developing methods for releasing GWAS data in a privacy-preserving way. Here, we develop an end-to-end differentially private method for solving regression problems with convex penalty functions and selecting the penalty parameters by cross-validation. In particular, we focus on penalized logistic regression with elastic-net regularization, a method widely used to in GWAS analyses to identify disease-causing genes. We show how a differentially private procedure for penalized logistic regression with elastic-net regularization can be applied to the analysis of GWAS data and evaluate our method’s performance.
AU - Yu, Fei
AU - Rybar, Michal
AU - Uhler, Caroline
AU - Fienberg, Stephen
ED - Domingo Ferrer, Josep
ID - 2047
T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
TI - Differentially-private logistic regression for detecting multiple-SNP association in GWAS databases
VL - 8744
ER -
TY - JOUR
AB - The flow instability and further transition to turbulence in a toroidal pipe (torus) with curvature ratio (tube-to-coiling diameter) 0.049 is investigated experimentally. The flow inside the toroidal pipe is driven by a steel sphere fitted to the inner pipe diameter. The sphere is moved with constant azimuthal velocity from outside the torus by a moving magnet. The experiment is designed to investigate curved pipe flow by optical measurement techniques. Using stereoscopic particle image velocimetry, laser Doppler velocimetry and pressure drop measurements, the flow is measured for Reynolds numbers ranging from 1000 to 15 000. Time- and space-resolved velocity fields are obtained and analysed. The steady axisymmetric basic flow is strongly influenced by centrifugal effects. On an increase of the Reynolds number we find a sequence of bifurcations. For Re=4075±2% a supercritical bifurcation to an oscillatory flow is found in which waves travel in the streamwise direction with a phase velocity slightly faster than the mean flow. The oscillatory flow is superseded by a presumably quasi-periodic flow at a further increase of the Reynolds number before turbulence sets in. The results are found to be compatible, in general, with earlier experimental and numerical investigations on transition to turbulence in helical and curved pipes. However, important aspects of the bifurcation scenario differ considerably.
AU - Kühnen, Jakob
AU - Holzner, Markus
AU - Hof, Björn
AU - Kuhlmann, Hendrik
ID - 2050
JF - Journal of Fluid Mechanics
TI - Experimental investigation of transitional flow in a toroidal pipe
VL - 738
ER -
TY - CONF
AB - A standard technique for solving the parameterized model checking problem is to reduce it to the classic model checking problem of finitely many finite-state systems. This work considers some of the theoretical power and limitations of this technique. We focus on concurrent systems in which processes communicate via pairwise rendezvous, as well as the special cases of disjunctive guards and token passing; specifications are expressed in indexed temporal logic without the next operator; and the underlying network topologies are generated by suitable Monadic Second Order Logic formulas and graph operations. First, we settle the exact computational complexity of the parameterized model checking problem for some of our concurrent systems, and establish new decidability results for others. Second, we consider the cases that model checking the parameterized system can be reduced to model checking some fixed number of processes, the number is known as a cutoff. We provide many cases for when such cutoffs can be computed, establish lower bounds on the size of such cutoffs, and identify cases where no cutoff exists. Third, we consider cases for which the parameterized system is equivalent to a single finite-state system (more precisely a Büchi word automaton), and establish tight bounds on the sizes of such automata.
AU - Aminof, Benjamin
AU - Kotek, Tomer
AU - Rubin, Sacha
AU - Spegni, Francesco
AU - Veith, Helmut
ED - Baldan, Paolo
ED - Gorla, Daniele
ID - 2052
T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
TI - Parameterized model checking of rendezvous systems
VL - 8704
ER -
TY - CONF
AB - In contrast to the usual understanding of probabilistic systems as stochastic processes, recently these systems have also been regarded as transformers of probabilities. In this paper, we give a natural definition of strong bisimulation for probabilistic systems corresponding to this view that treats probability distributions as first-class citizens. Our definition applies in the same way to discrete systems as well as to systems with uncountable state and action spaces. Several examples demonstrate that our definition refines the understanding of behavioural equivalences of probabilistic systems. In particular, it solves a longstanding open problem concerning the representation of memoryless continuous time by memoryfull continuous time. Finally, we give algorithms for computing this bisimulation not only for finite but also for classes of uncountably infinite systems.
AU - Hermanns, Holger
AU - Krčál, Jan
AU - Kretinsky, Jan
ED - Baldan, Paolo
ED - Gorla, Daniele
ID - 2053
T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
TI - Probabilistic bisimulation: Naturally on distributions
VL - 8704
ER -
TY - JOUR
AB - We consider a continuous-time Markov chain (CTMC) whose state space is partitioned into aggregates, and each aggregate is assigned a probability measure. A sufficient condition for defining a CTMC over the aggregates is presented as a variant of weak lumpability, which also characterizes that the measure over the original process can be recovered from that of the aggregated one. We show how the applicability of de-aggregation depends on the initial distribution. The application section is devoted to illustrate how the developed theory aids in reducing CTMC models of biochemical systems particularly in connection to protein-protein interactions. We assume that the model is written by a biologist in form of site-graph-rewrite rules. Site-graph-rewrite rules compactly express that, often, only a local context of a protein (instead of a full molecular species) needs to be in a certain configuration in order to trigger a reaction event. This observation leads to suitable aggregate Markov chains with smaller state spaces, thereby providing sufficient reduction in computational complexity. This is further exemplified in two case studies: simple unbounded polymerization and early EGFR/insulin crosstalk.
AU - Ganguly, Arnab
AU - Petrov, Tatjana
AU - Koeppl, Heinz
ID - 2056
IS - 3
JF - Journal of Mathematical Biology
TI - Markov chain aggregation and its applications to combinatorial reaction networks
VL - 69
ER -
TY - CONF
AB - In the past few years, a lot of attention has been devoted to multimedia indexing by fusing multimodal informations. Two kinds of fusion schemes are generally considered: The early fusion and the late fusion. We focus on late classifier fusion, where one combines the scores of each modality at the decision level. To tackle this problem, we investigate a recent and elegant well-founded quadratic program named MinCq coming from the machine learning PAC-Bayesian theory. MinCq looks for the weighted combination, over a set of real-valued functions seen as voters, leading to the lowest misclassification rate, while maximizing the voters’ diversity. We propose an extension of MinCq tailored to multimedia indexing. Our method is based on an order-preserving pairwise loss adapted to ranking that allows us to improve Mean Averaged Precision measure while taking into account the diversity of the voters that we want to fuse. We provide evidence that this method is naturally adapted to late fusion procedures and confirm the good behavior of our approach on the challenging PASCAL VOC’07 benchmark.
AU - Morvant, Emilie
AU - Habrard, Amaury
AU - Ayache, Stéphane
ID - 2057
T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
TI - Majority vote of diverse classifiers for late fusion
VL - 8621
ER -
TY - JOUR
AB - Plant embryogenesis is regulated by differential distribution of the plant hormone auxin. However, the cells establishing these gradients during microspore embryogenesis remain to be identified. For the first time, we describe, using the DR5 or DR5rev reporter gene systems, the GFP- and GUS-based auxin biosensors to monitor auxin during Brassica napus androgenesis at cellular resolution in the initial stages. Our study provides evidence that the distribution of auxin changes during embryo development and depends on the temperature-inducible in vitro culture conditions. For this, microspores (mcs) were induced to embryogenesis by heat treatment and then subjected to genetic modification via Agrobacterium tumefaciens. The duration of high temperature treatment had a significant influence on auxin distribution in isolated and in vitro-cultured microspores and on microspore-derived embryo development. In the “mild” heat-treated (1 day at 32 °C) mcs, auxin localized in a polar way already at the uni-nucleate microspore, which was critical for the initiation of embryos with suspensor-like structure. Assuming a mean mcs radius of 20 μm, endogenous auxin content in a single cell corresponded to concentration of 1.01 μM. In mcs subjected to a prolonged heat (5 days at 32 °C), although auxin concentration increased dozen times, auxin polarization was set up at a few-celled pro-embryos without suspensor. Those embryos were enclosed in the outer wall called the exine. The exine rupture was accompanied by the auxin gradient polarization. Relative quantitative estimation of auxin, using time-lapse imaging, revealed that primordia possess up to 1.3-fold higher amounts than those found in the root apices of transgenic MDEs in the presence of exogenous auxin. Our results show, for the first time, which concentration of endogenous auxin coincides with the first cell division and how the high temperature interplays with auxin, by what affects delay early establishing microspore polarity. Moreover, we present how the local auxin accumulation demonstrates the apical–basal axis formation of the androgenic embryo and directs the axiality of the adult haploid plant.
AU - Dubas, Ewa
AU - Moravčíková, Jana
AU - Libantová, Jana
AU - Matušíková, Ildikó
AU - Benková, Eva
AU - Zur, Iwona
AU - Krzewska, Monika
ID - 2059
IS - 5
JF - Protoplasma
TI - The influence of heat stress on auxin distribution in transgenic B napus microspores and microspore derived embryos
VL - 251
ER -
TY - JOUR
AB - Development of cambium and its activity is important for our knowledge of the mechanism of secondary growth. Arabidopsis thaliana emerges as a good model plant for such a kind of study. Thus, this paper reports on cellular events taking place in the interfascicular regions of inflorescence stems of A. thaliana, leading to the development of interfascicular cambium from differentiated interfascicular parenchyma cells (IPC). These events are as follows: appearance of auxin accumulation, PIN1 gene expression, polar PIN1 protein localization in the basal plasma membrane and periclinal divisions. Distribution of auxin was observed to be higher in differentiating into cambium parenchyma cells compared to cells within the pith and cortex. Expression of PIN1 in IPC was always preceded by auxin accumulation. Basal localization of PIN1 was already established in the cells prior to their periclinal division. These cellular events initiated within parenchyma cells adjacent to the vascular bundles and successively extended from that point towards the middle region of the interfascicular area, located between neighboring vascular bundles. The final consequence of which was the closure of the cambial ring within the stem. Changes in the chemical composition of IPC walls were also detected and included changes of pectic epitopes, xyloglucans (XG) and extensins rich in hydroxyproline (HRGPs). In summary, results presented in this paper describe interfascicular cambium ontogenesis in terms of successive cellular events in the interfascicular regions of inflorescence stems of Arabidopsis.
AU - Mazur, Ewa
AU - Kurczyñska, Ewa
AU - Friml, Jiří
ID - 2061
IS - 5
JF - Protoplasma
TI - Cellular events during interfascicular cambium ontogenesis in inflorescence stems of Arabidopsis
VL - 251
ER -
TY - JOUR
AB - The success story of fast-spiking, parvalbumin-positive (PV+) GABAergic interneurons (GABA, γ-aminobutyric acid) in the mammalian central nervous system is noteworthy. In 1995, the properties of these interneurons were completely unknown. Twenty years later, thanks to the massive use of subcellular patch-clamp techniques, simultaneous multiple-cell recording, optogenetics, in vivo measurements, and computational approaches, our knowledge about PV+ interneurons became more extensive than for several types of pyramidal neurons. These findings have implications beyond the “small world” of basic research on GABAergic cells. For example, the results provide a first proof of principle that neuroscientists might be able to close the gaps between the molecular, cellular, network, and behavioral levels, representing one of the main challenges at the present time. Furthermore, the results may form the basis for PV+ interneurons as therapeutic targets for brain disease in the future. However, much needs to be learned about the basic function of these interneurons before clinical neuroscientists will be able to use PV+ interneurons for therapeutic purposes.
AU - Hu, Hua
AU - Gan, Jian
AU - Jonas, Peter M
ID - 2062
IS - 6196
JF - Science
TI - Fast-spiking parvalbumin^+ GABAergic interneurons: From cellular design to microcircuit function
VL - 345
ER -
TY - JOUR
AB - Spin-wave theory is a key ingredient in our comprehension of quantum spin systems, and is used successfully for understanding a wide range of magnetic phenomena, including magnon condensation and stability of patterns in dipolar systems. Nevertheless, several decades of research failed to establish the validity of spin-wave theory rigorously, even for the simplest models of quantum spins. A rigorous justification of the method for the three-dimensional quantum Heisenberg ferromagnet at low temperatures is presented here. We derive sharp bounds on its free energy by combining a bosonic formulation of the model introduced by Holstein and Primakoff with probabilistic estimates and operator inequalities.
AU - Correggi, Michele
AU - Giuliani, Alessandro
AU - Seiringer, Robert
ID - 2029
IS - 2
JF - EPL
TI - Validity of spin-wave theory for the quantum Heisenberg model
VL - 108
ER -
TY - JOUR
AB - A puzzling property of synaptic transmission, originally established at the neuromuscular junction, is that the time course of transmitter release is independent of the extracellular Ca2+ concentration ([Ca2+]o), whereas the rate of release is highly [Ca2+]o-dependent. Here, we examine the time course of release at inhibitory basket cell-Purkinje cell synapses and show that it is independent of [Ca2+]o. Modeling of Ca2+-dependent transmitter release suggests that the invariant time course of release critically depends on tight coupling between Ca2+ channels and release sensors. Experiments with exogenous Ca2+ chelators reveal that channel-sensor coupling at basket cell-Purkinje cell synapses is very tight, with a mean distance of 10–20 nm. Thus, tight channel-sensor coupling provides a mechanistic explanation for the apparent [Ca2+]o independence of the time course of release.
AU - Arai, Itaru
AU - Jonas, Peter M
ID - 2031
JF - eLife
TI - Nanodomain coupling explains Ca^2+ independence of transmitter release time course at a fast central synapse
VL - 3
ER -
TY - JOUR
AB - As light-based control of fundamental signaling pathways is becoming a reality, the field of optogenetics is rapidly moving beyond neuroscience. We have recently developed receptor tyrosine kinases that are activated by light and control cell proliferation, epithelial–mesenchymal transition, and angiogenic sprouting—cell behaviors central to cancer progression.
AU - Inglés Prieto, Álvaro
AU - Gschaider-Reichhart, Eva
AU - Schelch, Karin
AU - Janovjak, Harald L
AU - Grusch, Michael
ID - 2032
IS - 4
JF - Molecular and Cellular Oncology
TI - The optogenetic promise for oncology: Episode I
VL - 1
ER -
TY - CONF
AB - The learning with privileged information setting has recently attracted a lot of attention within the machine learning community, as it allows the integration of additional knowledge into the training process of a classifier, even when this comes in the form of a data modality that is not available at test time. Here, we show that privileged information can naturally be treated as noise in the latent function of a Gaussian process classifier (GPC). That is, in contrast to the standard GPC setting, the latent function is not just a nuisance but a feature: it becomes a natural measure of confidence about the training data by modulating the slope of the GPC probit likelihood function. Extensive experiments on public datasets show that the proposed GPC method using privileged noise, called GPC+, improves over a standard GPC without privileged knowledge, and also over the current state-of-the-art SVM-based method, SVM+. Moreover, we show that advanced neural networks and deep learning methods can be compressed as privileged information.
AU - Hernandez Lobato, Daniel
AU - Sharmanska, Viktoriia
AU - Kersting, Kristian
AU - Lampert, Christoph
AU - Quadrianto, Novi
ID - 2033
IS - January
T2 - Advances in Neural Information Processing Systems
TI - Mind the nuisance: Gaussian process classification using privileged noise
VL - 1
ER -
TY - JOUR
AB - In rapidly changing environments, selection history may impact the dynamics of adaptation. Mutations selected in one environment may result in pleiotropic fitness trade-offs in subsequent novel environments, slowing the rates of adaptation. Epistatic interactions between mutations selected in sequential stressful environments may slow or accelerate subsequent rates of adaptation, depending on the nature of that interaction. We explored the dynamics of adaptation during sequential exposure to herbicides with different modes of action in Chlamydomonas reinhardtii. Evolution of resistance to two of the herbicides was largely independent of selection history. For carbetamide, previous adaptation to other herbicide modes of action positively impacted the likelihood of adaptation to this herbicide. Furthermore, while adaptation to all individual herbicides was associated with pleiotropic fitness costs in stress-free environments, we observed that accumulation of resistance mechanisms was accompanied by a reduction in overall fitness costs. We suggest that antagonistic epistasis may be a driving mechanism that enables populations to more readily adapt in novel environments. These findings highlight the potential for sequences of xenobiotics to facilitate the rapid evolution of multiple-drug and -pesticide resistance, as well as the potential for epistatic interactions between adaptive mutations to facilitate evolutionary rescue in rapidly changing environments.
AU - Lagator, Mato
AU - Colegrave, Nick
AU - Neve, Paul
ID - 2036
IS - 1794
JF - Proceedings of the Royal Society of London Series B Biological Sciences
TI - Selection history and epistatic interactions impact dynamics of adaptation to novel environmental stresses
VL - 281
ER -
TY - JOUR
AB - Understanding the evolution of dispersal is essential for understanding and predicting the dynamics of natural populations. Two main factors are known to influence dispersal evolution: spatio-temporal variation in the environment and relatedness between individuals. However, the relation between these factors is still poorly understood, and they are usually treated separately. In this article, I present a theoretical framework that contains and connects effects of both environmental variation and relatedness, and reproduces and extends their known features. Spatial habitat variation selects for balanced dispersal strategies, whereby the population is kept at an ideal free distribution. Within this class of dispersal strategies, I explain how increased dispersal is promoted by perturbations to the dispersal type frequencies. An explicit formula shows the magnitude of the selective advantage of increased dispersal in terms of the spatial variability in the frequencies of the different dispersal strategies present. These variances are capable of capturing various sources of stochasticity and hence establish a common scale for their effects on the evolution of dispersal. The results furthermore indicate an alternative approach to identifying effects of relatedness on dispersal evolution.
AU - Novak, Sebastian
ID - 2023
IS - 24
JF - Ecology and Evolution
TI - Habitat heterogeneities versus spatial type frequency variances as driving forces of dispersal evolution
VL - 4
ER -
TY - JOUR
AB - The yeast Rab5 homologue, Vps21p, is known to be involved both in the vacuolar protein sorting (VPS) pathway from the trans-Golgi network to the vacuole, and in the endocytic pathway from the plasma membrane to the vacuole. However, the intracellular location at which these two pathways converge remains unclear. In addition, the endocytic pathway is not completely blocked in yeast cells lacking all Rab5 genes, suggesting the existence of an unidentified route that bypasses the Rab5-dependent endocytic pathway. Here we show that convergence of the endocytic and VPS pathways occurs upstream of the requirement for Vps21p in these pathways. We also identify a previously unidentified endocytic pathway mediated by the AP-3 complex. Importantly, the AP-3-mediated pathway appears mostly intact in Rab5-disrupted cells, and thus works as an alternative route to the vacuole/lysosome. We propose that the endocytic traffic branches into two routes to reach the vacuole: a Rab5-dependent VPS pathway and a Rab5-independent AP-3-mediated pathway.
AU - Toshima, Junko
AU - Nishinoaki, Show
AU - Sato, Yoshifumi
AU - Yamamoto, Wataru
AU - Furukawa, Daiki
AU - Siekhaus, Daria E
AU - Sawaguchi, Akira
AU - Toshima, Jiro
ID - 2024
JF - Nature Communications
TI - Bifurcation of the endocytic pathway into Rab5-dependent and -independent transport to the vacuole
VL - 5
ER -
TY - CONF
AB - We present a tool for translating LTL formulae into deterministic ω-automata. It is the first tool that covers the whole LTL that does not use Safra’s determinization or any of its variants. This leads to smaller automata. There are several outputs of the tool: firstly, deterministic Rabin automata, which are the standard input for probabilistic model checking, e.g. for the probabilistic model-checker PRISM; secondly, deterministic generalized Rabin automata, which can also be used for probabilistic model checking and are sometimes by orders of magnitude smaller. We also link our tool to PRISM and show that this leads to a significant speed-up of probabilistic LTL model checking, especially with the generalized Rabin automata.
AU - Komárková, Zuzana
AU - Kretinsky, Jan
ED - Cassez, Franck
ED - Raskin, Jean-François
ID - 2026
T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
TI - Rabinizer 3: Safraless translation of ltl to small deterministic automata
VL - 8837
ER -
TY - CONF
AB - We present a general framework for applying machine-learning algorithms to the verification of Markov decision processes (MDPs). The primary goal of these techniques is to improve performance by avoiding an exhaustive exploration of the state space. Our framework focuses on probabilistic reachability, which is a core property for verification, and is illustrated through two distinct instantiations. The first assumes that full knowledge of the MDP is available, and performs a heuristic-driven partial exploration of the model, yielding precise lower and upper bounds on the required probability. The second tackles the case where we may only sample the MDP, and yields probabilistic guarantees, again in terms of both the lower and upper bounds, which provides efficient stopping criteria for the approximation. The latter is the first extension of statistical model checking for unbounded properties inMDPs. In contrast with other related techniques, our approach is not restricted to time-bounded (finite-horizon) or discounted properties, nor does it assume any particular properties of the MDP. We also show how our methods extend to LTL objectives. We present experimental results showing the performance of our framework on several examples.
AU - Brázdil, Tomáš
AU - Chatterjee, Krishnendu
AU - Chmelik, Martin
AU - Forejt, Vojtěch
AU - Kretinsky, Jan
AU - Kwiatkowska, Marta
AU - Parker, David
AU - Ujma, Mateusz
ED - Cassez, Franck
ED - Raskin, Jean-François
ID - 2027
T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
TI - Verification of markov decision processes using learning algorithms
VL - 8837
ER -
TY - JOUR
AB - We examined the synaptic structure, quantity, and distribution of α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA)- and N-methyl-D-aspartate (NMDA)-type glutamate receptors (AMPARs and NMDARs, respectively) in rat cochlear nuclei by a highly sensitive freeze-fracture replica labeling technique. Four excitatory synapses formed by two distinct inputs, auditory nerve (AN) and parallel fibers (PF), on different cell types were analyzed. These excitatory synapse types included AN synapses on bushy cells (AN-BC synapses) and fusiform cells (AN-FC synapses) and PF synapses on FC (PF-FC synapses) and cartwheel cell spines (PF-CwC synapses). Immunogold labeling revealed differences in synaptic structure as well as AMPAR and NMDAR number and/or density in both AN and PF synapses, indicating a target-dependent organization. The immunogold receptor labeling also identified differences in the synaptic organization of FCs based on AN or PF connections, indicating an input-dependent organization in FCs. Among the four excitatory synapse types, the AN-BC synapses were the smallest and had the most densely packed intramembrane particles (IMPs), whereas the PF-CwC synapses were the largest and had sparsely packed IMPs. All four synapse types showed positive correlations between the IMP-cluster area and the AMPAR number, indicating a common intrasynapse-type relationship for glutamatergic synapses. Immunogold particles for AMPARs were distributed over the entire area of individual AN synapses; PF synapses often showed synaptic areas devoid of labeling. The gold-labeling for NMDARs occurred in a mosaic fashion, with less positive correlations between the IMP-cluster area and the NMDAR number. Our observations reveal target- and input-dependent features in the structure, number, and organization of AMPARs and NMDARs in AN and PF synapses.
AU - Rubio, Maía
AU - Fukazawa, Yugo
AU - Kamasawa, Naomi
AU - Clarkson, Cheryl
AU - Molnár, Elek
AU - Shigemoto, Ryuichi
ID - 2064
IS - 18
JF - Journal of Comparative Neurology
TI - Target- and input-dependent organization of AMPA and NMDA receptors in synaptic connections of the cochlear nucleus
VL - 522
ER -
TY - JOUR
AB - Pathogens may gain a fitness advantage through manipulation of the behaviour of their hosts. Likewise, host behavioural changes can be a defence mechanism, counteracting the impact of pathogens on host fitness. We apply harmonic radar technology to characterize the impact of an emerging pathogen - Nosema ceranae (Microsporidia) - on honeybee (Apis mellifera) flight and orientation performance in the field. Honeybees are the most important commercial pollinators. Emerging diseases have been proposed to play a prominent role in colony decline, partly through sub-lethal behavioural manipulation of their hosts. We found that homing success was significantly reduced in diseased (65.8%) versus healthy foragers (92.5%). Although lost bees had significantly reduced continuous flight times and prolonged resting times, other flight characteristics and navigational abilities showed no significant difference between infected and non-infected bees. Our results suggest that infected bees express normal flight characteristics but are constrained in their homing ability, potentially compromising the colony by reducing its resource inputs, but also counteracting the intra-colony spread of infection. We provide the first high-resolution analysis of sub-lethal effects of an emerging disease on insect flight behaviour. The potential causes and the implications for both host and parasite are discussed.
AU - Wolf, Stephan
AU - Mcmahon, Dino
AU - Lim, Ka
AU - Pull, Christopher
AU - Clark, Suzanne
AU - Paxton, Robert
AU - Osborne, Juliet
ID - 2086
IS - 8
JF - PLoS One
TI - So near and yet so far: Harmonic radar reveals reduced homing ability of Nosema infected honeybees
VL - 9
ER -
TY - JOUR
AB - Understanding the effects of sex and migration on adaptation to novel environments remains a key problem in evolutionary biology. Using a single-cell alga Chlamydomonas reinhardtii, we investigated how sex and migration affected rates of evolutionary rescue in a sink environment, and subsequent changes in fitness following evolutionary rescue. We show that sex and migration affect both the rate of evolutionary rescue and subsequent adaptation. However, their combined effects change as the populations adapt to a sink habitat. Both sex and migration independently increased rates of evolutionary rescue, but the effect of sex on subsequent fitness improvements, following initial rescue, changed with migration, as sex was beneficial in the absence of migration but constraining adaptation when combined with migration. These results suggest that sex and migration are beneficial during the initial stages of adaptation, but can become detrimental as the population adapts to its environment.
AU - Lagator, Mato
AU - Morgan, Andrew
AU - Neve, Paul
AU - Colegrave, Nick
ID - 2083
IS - 8
JF - Evolution
TI - Role of sex and migration in adaptation to sink environments
VL - 68
ER -
TY - CONF
AB - We define a simple, explicit map sending a morphism f : M → N of pointwise finite dimensional persistence modules to a matching between the barcodes of M and N. Our main result is that, in a precise sense, the quality of this matching is tightly controlled by the lengths of the longest intervals in the barcodes of ker f and coker f . As an immediate corollary, we obtain a new proof of the algebraic stability theorem for persistence barcodes [5, 9], a fundamental result in the theory of persistent homology. In contrast to previous proofs, ours shows explicitly how a δ-interleaving morphism between two persistence modules induces a δ-matching between the barcodes of the two modules. Our main result also specializes to a structure theorem for submodules and quotients of persistence modules. Copyright is held by the owner/author(s).
AU - Bauer, Ulrich
AU - Lesnick, Michael
ID - 2153
T2 - Proceedings of the Annual Symposium on Computational Geometry
TI - Induced matchings of barcodes and the algebraic stability of persistence
ER -
TY - JOUR
AB - A result of Boros and Füredi (d = 2) and of Bárány (arbitrary d) asserts that for every d there exists cd > 0 such that for every n-point set P ⊂ ℝd, some point of ℝd is covered by at least (Formula presented.) of the d-simplices spanned by the points of P. The largest possible value of cd has been the subject of ongoing research. Recently Gromov improved the existing lower bounds considerably by introducing a new, topological proof method. We provide an exposition of the combinatorial component of Gromov's approach, in terms accessible to combinatorialists and discrete geometers, and we investigate the limits of his method. In particular, we give tighter bounds on the cofilling profiles for the (n - 1)-simplex. These bounds yield a minor improvement over Gromov's lower bounds on cd for large d, but they also show that the room for further improvement through the cofilling profiles alone is quite small. We also prove a slightly better lower bound for c3 by an approach using an additional structure besides the cofilling profiles. We formulate a combinatorial extremal problem whose solution might perhaps lead to a tight lower bound for cd.
AU - Matoušek, Jiří
AU - Wagner, Uli
ID - 2154
IS - 1
JF - Discrete & Computational Geometry
TI - On Gromov's method of selecting heavily covered points
VL - 52
ER -
TY - CONF
AB - Given a finite set of points in Rn and a positive radius, we study the Čech, Delaunay-Čech, alpha, and wrap complexes as instances of a generalized discrete Morse theory. We prove that the latter three complexes are simple-homotopy equivalent. Our results have applications in topological data analysis and in the reconstruction of shapes from sampled data. Copyright is held by the owner/author(s).
AU - Bauer, Ulrich
AU - Edelsbrunner, Herbert
ID - 2155
T2 - Proceedings of the Annual Symposium on Computational Geometry
TI - The morse theory of Čech and Delaunay filtrations
ER -
TY - CONF
AB - We propose a metric for Reeb graphs, called the functional distortion distance. Under this distance, the Reeb graph is stable against small changes of input functions. At the same time, it remains discriminative at differentiating input functions. In particular, the main result is that the functional distortion distance between two Reeb graphs is bounded from below by the bottleneck distance between both the ordinary and extended persistence diagrams for appropriate dimensions. As an application of our results, we analyze a natural simplification scheme for Reeb graphs, and show that persistent features in Reeb graph remains persistent under simplification. Understanding the stability of important features of the Reeb graph under simplification is an interesting problem on its own right, and critical to the practical usage of Reeb graphs. Copyright is held by the owner/author(s).
AU - Bauer, Ulrich
AU - Ge, Xiaoyin
AU - Wang, Yusu
ID - 2156
T2 - Proceedings of the Annual Symposium on Computational Geometry
TI - Measuring distance between Reeb graphs
ER -
TY - CONF
AB - Motivated by topological Tverberg-type problems, we consider multiple (double, triple, and higher multiplicity) selfintersection points of maps from finite simplicial complexes (compact polyhedra) into ℝd and study conditions under which such multiple points can be eliminated. The most classical case is that of embeddings (i.e., maps without double points) of a κ-dimensional complex K into ℝ2κ. For this problem, the work of van Kampen, Shapiro, and Wu provides an efficiently testable necessary condition for embeddability (namely, vanishing of the van Kampen ob-struction). For κ ≥ 3, the condition is also sufficient, and yields a polynomial-time algorithm for deciding embeddability: One starts with an arbitrary map f : K→ℝ2κ, which generically has finitely many double points; if k ≥ 3 and if the obstruction vanishes then one can successively remove these double points by local modifications of the map f. One of the main tools is the famous Whitney trick that permits eliminating pairs of double points of opposite intersection sign. We are interested in generalizing this approach to intersection points of higher multiplicity. We call a point y 2 ℝd an r-fold Tverberg point of a map f : Kκ →ℝd if y lies in the intersection f(σ1)∩. ∩f(σr) of the images of r pairwise disjoint simplices of K. The analogue of (non-)embeddability that we study is the problem Tverbergκ r→d: Given a κ-dimensional complex K, does it satisfy a Tverberg-type theorem with parameters r and d, i.e., does every map f : K κ → ℝd have an r-fold Tverberg point? Here, we show that for fixed r, κ and d of the form d = rm and k = (r-1)m, m ≥ 3, there is a polynomial-time algorithm for deciding this (based on the vanishing of a cohomological obstruction, as in the case of embeddings). Our main tool is an r-fold analogue of the Whitney trick: Given r pairwise disjoint simplices of K such that the intersection of their images contains two r-fold Tverberg points y+ and y- of opposite intersection sign, we can eliminate y+ and y- by a local isotopy of f. In a subsequent paper, we plan to develop this further and present a generalization of the classical Haeiger-Weber Theorem (which yields a necessary and sufficient condition for embeddability of κ-complexes into ℝd for a wider range of dimensions) to intersection points of higher multiplicity.
AU - Mabillard, Isaac
AU - Wagner, Uli
ID - 2159
T2 - Proceedings of the Annual Symposium on Computational Geometry
TI - Eliminating Tverberg points, I. An analogue of the Whitney trick
ER -
TY - CONF
AB - Transfer learning has received a lot of attention in the machine learning community over the last years, and several effective algorithms have been developed. However, relatively little is known about their theoretical properties, especially in the setting of lifelong learning, where the goal is to transfer information to tasks for which no data have been observed so far. In this work we study lifelong learning from a theoretical perspective. Our main result is a PAC-Bayesian generalization bound that offers a unified view on existing paradigms for transfer learning, such as the transfer of parameters or the transfer of low-dimensional representations. We also use the bound to derive two principled lifelong learning algorithms, and we show that these yield results comparable with existing methods.
AU - Pentina, Anastasia
AU - Lampert, Christoph
ED - Xing, Eric
ED - Jebara, Tony
ID - 2160
TI - A PAC-Bayesian bound for Lifelong Learning
VL - 32
ER -
TY - JOUR
AB - Repeated pathogen exposure is a common threat in colonies of social insects, posing selection pressures on colony members to respond with improved disease-defense performance. We here tested whether experience gained by repeated tending of low-level fungus-exposed (Metarhizium robertsii) larvae may alter the performance of sanitary brood care in the clonal ant, Platythyrea punctata. We trained ants individually over nine consecutive trials to either sham-treated or fungus-exposed larvae. We then compared the larval grooming behavior of naive and trained ants and measured how effectively they removed infectious fungal conidiospores from the fungus-exposed larvae. We found that the ants changed the duration of larval grooming in response to both, larval treatment and their level of experience: (1) sham-treated larvae received longer grooming than the fungus-exposed larvae and (2) trained ants performed less self-grooming but longer larval grooming than naive ants, which was true for both, ants trained to fungus-exposed and also to sham-treated larvae. Ants that groomed the fungus-exposed larvae for longer periods removed a higher number of fungal conidiospores from the surface of the fungus-exposed larvae. As experienced ants performed longer larval grooming, they were more effective in fungal removal, thus making them better caretakers under pathogen attack of the colony. By studying this clonal ant, we can thus conclude that even in the absence of genetic variation between colony members, differences in experience levels of brood care may affect performance of sanitary brood care in social insects.
AU - Westhus, Claudia
AU - Ugelvig, Line V
AU - Tourdot, Edouard
AU - Heinze, Jürgen
AU - Doums, Claudie
AU - Cremer, Sylvia
ID - 2161
IS - 10
JF - Behavioral Ecology and Sociobiology
TI - Increased grooming after repeated brood care provides sanitary benefits in a clonal ant
VL - 68
ER -
TY - JOUR
AB - Directional guidance of migrating cells is relatively well explored in the reductionist setting of cell culture experiments. Here spatial gradients of chemical cues as well as gradients of mechanical substrate characteristics prove sufficient to attract single cells as well as their collectives. How such gradients present and act in the context of an organism is far less clear. Here we review recent advances in understanding how guidance cues emerge and operate in the physiological context.
AU - Majumdar, Ritankar
AU - Sixt, Michael K
AU - Parent, Carole
ID - 2158
IS - 1
JF - Current Opinion in Cell Biology
TI - New paradigms in the establishment and maintenance of gradients during directed cell migration
VL - 30
ER -
TY - JOUR
AB - Many species have an essentially continuous distribution in space, in which there are no natural divisions between randomly mating subpopulations. Yet, the standard approach to modelling these populations is to impose an arbitrary grid of demes, adjusting deme sizes and migration rates in an attempt to capture the important features of the population. Such indirect methods are required because of the failure of the classical models of isolation by distance, which have been shown to have major technical flaws. A recently introduced model of extinction and recolonisation in two dimensions solves these technical problems, and provides a rigorous technical foundation for the study of populations evolving in a spatial continuum. The coalescent process for this model is simply stated, but direct simulation is very inefficient for large neighbourhood sizes. We present efficient and exact algorithms to simulate this coalescent process for arbitrary sample sizes and numbers of loci, and analyse these algorithms in detail.
AU - Kelleher, Jerome
AU - Etheridge, Alison
AU - Barton, Nicholas H
ID - 2168
JF - Theoretical Population Biology
TI - Coalescent simulation in continuous space: Algorithms for large neighbourhood size
VL - 95
ER -
TY - JOUR
AU - Barton, Nicholas H
AU - Novak, Sebastian
AU - Paixao, Tiago
ID - 2169
IS - 29
JF - PNAS
TI - Diverse forms of selection in evolution and computer science
VL - 111
ER -
TY - JOUR
AB - Short-read sequencing technologies have in principle made it feasible to draw detailed inferences about the recent history of any organism. In practice, however, this remains challenging due to the difficulty of genome assembly in most organisms and the lack of statistical methods powerful enough to discriminate between recent, nonequilibrium histories. We address both the assembly and inference challenges. We develop a bioinformatic pipeline for generating outgroup-rooted alignments of orthologous sequence blocks from de novo low-coverage short-read data for a small number of genomes, and show how such sequence blocks can be used to fit explicit models of population divergence and admixture in a likelihood framework. To illustrate our approach, we reconstruct the Pleistocene history of an oak-feeding insect (the oak gallwasp Biorhiza pallida), which, in common with many other taxa, was restricted during Pleistocene ice ages to a longitudinal series of southern refugia spanning the Western Palaearctic. Our analysis of sequence blocks sampled from a single genome from each of three major glacial refugia reveals support for an unexpected history dominated by recent admixture. Despite the fact that 80% of the genome is affected by admixture during the last glacial cycle, we are able to infer the deeper divergence history of these populations. These inferences are robust to variation in block length, mutation model and the sampling location of individual genomes within refugia. This combination of de novo assembly and numerical likelihood calculation provides a powerful framework for estimating recent population history that can be applied to any organism without the need for prior genetic resources.
AU - Hearn, Jack
AU - Stone, Graham
AU - Bunnefeld, Lynsey
AU - Nicholls, James
AU - Barton, Nicholas H
AU - Lohse, Konrad
ID - 2170
IS - 1
JF - Molecular Ecology
TI - Likelihood-based inference of population history from low-coverage de novo genome assemblies
VL - 23
ER -
TY - CONF
AB - We present LS-CRF, a new method for training cyclic Conditional Random Fields (CRFs) from large datasets that is inspired by classical closed-form expressions for the maximum likelihood parameters of a generative graphical model with tree topology. Training a CRF with LS-CRF requires only solving a set of independent regression problems, each of which can be solved efficiently in closed form or by an iterative solver. This makes LS-CRF orders of magnitude faster than classical CRF training based on probabilistic inference, and at the same time more flexible and easier to implement than other approximate techniques, such as pseudolikelihood or piecewise training. We apply LS-CRF to the task of semantic image segmentation, showing that it achieves on par accuracy to other training techniques at higher speed, thereby allowing efficient CRF training from very large training sets. For example, training a linearly parameterized pairwise CRF on 150,000 images requires less than one hour on a modern workstation.
AU - Kolesnikov, Alexander
AU - Guillaumin, Matthieu
AU - Ferrari, Vittorio
AU - Lampert, Christoph
ED - Fleet, David
ED - Pajdla, Tomas
ED - Schiele, Bernt
ED - Tuytelaars, Tinne
ID - 2171
IS - PART 3
T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
TI - Closed-form approximate CRF training for scalable image segmentation
VL - 8691
ER -
TY - CONF
AB - Fisher Kernels and Deep Learning were two developments with significant impact on large-scale object categorization in the last years. Both approaches were shown to achieve state-of-the-art results on large-scale object categorization datasets, such as ImageNet. Conceptually, however, they are perceived as very different and it is not uncommon for heated debates to spring up when advocates of both paradigms meet at conferences or workshops. In this work, we emphasize the similarities between both architectures rather than their differences and we argue that such a unified view allows us to transfer ideas from one domain to the other. As a concrete example we introduce a method for learning a support vector machine classifier with Fisher kernel at the same time as a task-specific data representation. We reinterpret the setting as a multi-layer feed forward network. Its final layer is the classifier, parameterized by a weight vector, and the two previous layers compute Fisher vectors, parameterized by the coefficients of a Gaussian mixture model. We introduce a gradient descent based learning algorithm that, in contrast to other feature learning techniques, is not just derived from intuition or biological analogy, but has a theoretical justification in the framework of statistical learning theory. Our experiments show that the new training procedure leads to significant improvements in classification accuracy while preserving the modularity and geometric interpretability of a support vector machine setup.
AU - Sydorov, Vladyslav
AU - Sakurada, Mayu
AU - Lampert, Christoph
ID - 2172
T2 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
TI - Deep Fisher Kernels – End to end learning of the Fisher Kernel GMM parameters
ER -
TY - CONF
AB - In this work we introduce a new approach to co-classification, i.e. the task of jointly classifying multiple, otherwise independent, data samples. The method we present, named CoConut, is based on the idea of adding a regularizer in the label space to encode certain priors on the resulting labelings. A regularizer that encourages labelings that are smooth across the test set, for instance, can be seen as a test-time variant of the cluster assumption, which has been proven useful at training time in semi-supervised learning. A regularizer that introduces a preference for certain class proportions can be regarded as a prior distribution on the class labels. CoConut can build on existing classifiers without making any assumptions on how they were obtained and without the need to re-train them. The use of a regularizer adds a new level of flexibility. It allows the integration of potentially new information at test time, even in other modalities than what the classifiers were trained on. We evaluate our framework on six datasets, reporting a clear performance gain in classification accuracy compared to the standard classification setup that predicts labels for each test sample separately.
AU - Khamis, Sameh
AU - Lampert, Christoph
ID - 2173
T2 - Proceedings of the British Machine Vision Conference 2014
TI - CoConut: Co-classification with output space regularization
ER -
TY - JOUR
AB - When polygenic traits are under stabilizing selection, many different combinations of alleles allow close adaptation to the optimum. If alleles have equal effects, all combinations that result in the same deviation from the optimum are equivalent. Furthermore, the genetic variance that is maintained by mutation-selection balance is 2μ/S per locus, where μ is the mutation rate and S the strength of stabilizing selection. In reality, alleles vary in their effects, making the fitness landscape asymmetric and complicating analysis of the equilibria. We show that that the resulting genetic variance depends on the fraction of alleles near fixation, which contribute by 2μ/S, and on the total mutational effects of alleles that are at intermediate frequency. The inpplayfi between stabilizing selection and mutation leads to a sharp transition: alleles with effects smaller than a threshold value of 2 remain polymorphic, whereas those with larger effects are fixed. The genetic load in equilibrium is less than for traits of equal effects, and the fitness equilibria are more similar. We find p the optimum is displaced, alleles with effects close to the threshold value sweep first, and their rate of increase is bounded by Long-term response leads in general to well-adapted traits, unlike the case of equal effects that often end up at a suboptimal fitness peak. However, the particular peaks to which the populations converge are extremely sensitive to the initial states and to the speed of the shift of the optimum trait value.
AU - De Vladar, Harold
AU - Barton, Nicholas H
ID - 2174
IS - 2
JF - Genetics
TI - Stability and response of polygenic traits to stabilizing selection and mutation
VL - 197
ER -
TY - JOUR
AB - The cerebral cortex, the seat of our cognitive abilities, is composed of an intricate network of billions of excitatory projection and inhibitory interneurons. Postmitotic cortical neurons are generated by a diverse set of neural stem cell progenitors within dedicated zones and defined periods of neurogenesis during embryonic development. Disruptions in neurogenesis can lead to alterations in the neuronal cytoarchitecture, which is thought to represent a major underlying cause for several neurological disorders, including microcephaly, autism and epilepsy. Although a number of signaling pathways regulating neurogenesis have been described, the precise cellular and molecular mechanisms regulating the functional neural stem cell properties in cortical neurogenesis remain unclear. Here, we discuss the most up-to-date strategies to monitor the fundamental mechanistic parameters of neuronal progenitor proliferation, and recent advances deciphering the logic and dynamics of neurogenesis.
AU - Postiglione, Maria P
AU - Hippenmeyer, Simon
ID - 2175
IS - 3
JF - Future Neurology
TI - Monitoring neurogenesis in the cerebral cortex: an update
VL - 9
ER -
TY - JOUR
AB - Electron microscopy (EM) allows for the simultaneous visualization of all tissue components at high resolution. However, the extent to which conventional aldehyde fixation and ethanol dehydration of the tissue alter the fine structure of cells and organelles, thereby preventing detection of subtle structural changes induced by an experiment, has remained an issue. Attempts have been made to rapidly freeze tissue to preserve native ultrastructure. Shock-freezing of living tissue under high pressure (high-pressure freezing, HPF) followed by cryosubstitution of the tissue water avoids aldehyde fixation and dehydration in ethanol; the tissue water is immobilized in â ̂1/450 ms, and a close-to-native fine structure of cells, organelles and molecules is preserved. Here we describe a protocol for HPF that is useful to monitor ultrastructural changes associated with functional changes at synapses in the brain but can be applied to many other tissues as well. The procedure requires a high-pressure freezer and takes a minimum of 7 d but can be paused at several points.
AU - Studer, Daniel
AU - Zhao, Shanting
AU - Chai, Xuejun
AU - Jonas, Peter M
AU - Graber, Werner
AU - Nestel, Sigrun
AU - Frotscher, Michael
ID - 2176
IS - 6
JF - Nature Protocols
TI - Capture of activity-induced ultrastructural changes at synapses by high-pressure freezing of brain tissue
VL - 9
ER -
TY - CONF
AB - We give evidence for the difficulty of computing Betti numbers of simplicial complexes over a finite field. We do this by reducing the rank computation for sparse matrices with to non-zero entries to computing Betti numbers of simplicial complexes consisting of at most a constant times to simplices. Together with the known reduction in the other direction, this implies that the two problems have the same computational complexity.
AU - Edelsbrunner, Herbert
AU - Parsa, Salman
ID - 2177
T2 - Proceedings of the Annual ACM-SIAM Symposium on Discrete Algorithms
TI - On the computational complexity of betti numbers reductions from matrix rank
ER -
TY - JOUR
AB - We consider the three-state toric homogeneous Markov chain model (THMC) without loops and initial parameters. At time T, the size of the design matrix is 6 × 3 · 2T-1 and the convex hull of its columns is the model polytope. We study the behavior of this polytope for T ≥ 3 and we show that it is defined by 24 facets for all T ≥ 5. Moreover, we give a complete description of these facets. From this, we deduce that the toric ideal associated with the design matrix is generated by binomials of degree at most 6. Our proof is based on a result due to Sturmfels, who gave a bound on the degree of the generators of a toric ideal, provided the normality of the corresponding toric variety. In our setting, we established the normality of the toric variety associated to the THMC model by studying the geometric properties of the model polytope.
AU - Haws, David
AU - Martin Del Campo Sanchez, Abraham
AU - Takemura, Akimichi
AU - Yoshida, Ruriko
ID - 2178
IS - 1
JF - Beitrage zur Algebra und Geometrie
TI - Markov degree of the three-state toric homogeneous Markov chain model
VL - 55
ER -
TY - JOUR
AB - We extend the proof of the local semicircle law for generalized Wigner matrices given in MR3068390 to the case when the matrix of variances has an eigenvalue -1. In particular, this result provides a short proof of the optimal local Marchenko-Pastur law at the hard edge (i.e. around zero) for sample covariance matrices X*X, where the variances of the entries of X may vary.
AU - Ajanki, Oskari H
AU - Erdös, László
AU - Krüger, Torben H
ID - 2179
JF - Electronic Communications in Probability
TI - Local semicircle law with imprimitive variance matrix
VL - 19
ER -
TY - JOUR
AB - Weighted majority votes allow one to combine the output of several classifiers or voters. MinCq is a recent algorithm for optimizing the weight of each voter based on the minimization of a theoretical bound over the risk of the vote with elegant PAC-Bayesian generalization guarantees. However, while it has demonstrated good performance when combining weak classifiers, MinCq cannot make use of the useful a priori knowledge that one may have when using a mixture of weak and strong voters. In this paper, we propose P-MinCq, an extension of MinCq that can incorporate such knowledge in the form of a constraint over the distribution of the weights, along with general proofs of convergence that stand in the sample compression setting for data-dependent voters. The approach is applied to a vote of k-NN classifiers with a specific modeling of the voters' performance. P-MinCq significantly outperforms the classic k-NN classifier, a symmetric NN and MinCq using the same voters. We show that it is also competitive with LMNN, a popular metric learning algorithm, and that combining both approaches further reduces the error.
AU - Bellet, Aurélien
AU - Habrard, Amaury
AU - Morvant, Emilie
AU - Sebban, Marc
ID - 2180
IS - 1-2
JF - Machine Learning
TI - Learning a priori constrained weighted majority votes
VL - 97
ER -
TY - JOUR
AB - Given topological spaces X,Y, a fundamental problem of algebraic topology is understanding the structure of all continuous maps X→ Y. We consider a computational version, where X,Y are given as finite simplicial complexes, and the goal is to compute [X,Y], that is, all homotopy classes of suchmaps.We solve this problem in the stable range, where for some d ≥ 2, we have dim X ≤ 2d-2 and Y is (d-1)-connected; in particular, Y can be the d-dimensional sphere Sd. The algorithm combines classical tools and ideas from homotopy theory (obstruction theory, Postnikov systems, and simplicial sets) with algorithmic tools from effective algebraic topology (locally effective simplicial sets and objects with effective homology). In contrast, [X,Y] is known to be uncomputable for general X,Y, since for X = S1 it includes a well known undecidable problem: testing triviality of the fundamental group of Y. In follow-up papers, the algorithm is shown to run in polynomial time for d fixed, and extended to other problems, such as the extension problem, where we are given a subspace A ⊂ X and a map A→ Y and ask whether it extends to a map X → Y, or computing the Z2-index-everything in the stable range. Outside the stable range, the extension problem is undecidable.
AU - Čadek, Martin
AU - Krcál, Marek
AU - Matoušek, Jiří
AU - Sergeraert, Francis
AU - Vokřínek, Lukáš
AU - Wagner, Uli
ID - 2184
IS - 3
JF - Journal of the ACM
TI - Computing all maps into a sphere
VL - 61
ER -
TY - CONF
AB - We revisit the classical problem of converting an imperfect source of randomness into a usable cryptographic key. Assume that we have some cryptographic application P that expects a uniformly random m-bit key R and ensures that the best attack (in some complexity class) against P(R) has success probability at most δ. Our goal is to design a key-derivation function (KDF) h that converts any random source X of min-entropy k into a sufficiently "good" key h(X), guaranteeing that P(h(X)) has comparable security δ′ which is 'close' to δ. Seeded randomness extractors provide a generic way to solve this problem for all applications P, with resulting security δ′ = O(δ), provided that we start with entropy k ≥ m + 2 log (1/δ) - O(1). By a result of Radhakrishnan and Ta-Shma, this bound on k (called the "RT-bound") is also known to be tight in general. Unfortunately, in many situations the loss of 2 log (1/δ) bits of entropy is unacceptable. This motivates the study KDFs with less entropy waste by placing some restrictions on the source X or the application P. In this work we obtain the following new positive and negative results in this regard: - Efficient samplability of the source X does not help beat the RT-bound for general applications. This resolves the SRT (samplable RT) conjecture of Dachman-Soled et al. [DGKM12] in the affirmative, and also shows that the existence of computationally-secure extractors beating the RT-bound implies the existence of one-way functions. - We continue in the line of work initiated by Barak et al. [BDK+11] and construct new information-theoretic KDFs which beat the RT-bound for large but restricted classes of applications. Specifically, we design efficient KDFs that work for all unpredictability applications P (e.g., signatures, MACs, one-way functions, etc.) and can either: (1) extract all of the entropy k = m with a very modest security loss δ′ = O(δ·log (1/δ)), or alternatively, (2) achieve essentially optimal security δ′ = O(δ) with a very modest entropy loss k ≥ m + loglog (1/δ). In comparison, the best prior results from [BDK+11] for this class of applications would only guarantee δ′ = O(√δ) when k = m, and would need k ≥ m + log (1/δ) to get δ′ = O(δ). - The weaker bounds of [BDK+11] hold for a larger class of so-called "square- friendly" applications (which includes all unpredictability, but also some important indistinguishability, applications). Unfortunately, we show that these weaker bounds are tight for the larger class of applications. - We abstract out a clean, information-theoretic notion of (k,δ,δ′)- unpredictability extractors, which guarantee "induced" security δ′ for any δ-secure unpredictability application P, and characterize the parameters achievable for such unpredictability extractors. Of independent interest, we also relate this notion to the previously-known notion of (min-entropy) condensers, and improve the state-of-the-art parameters for such condensers.
AU - Dodis, Yevgeniy
AU - Pietrzak, Krzysztof Z
AU - Wichs, Daniel
ED - Nguyen, Phong
ED - Oswald, Elisabeth
ID - 2185
TI - Key derivation without entropy waste
VL - 8441
ER -
TY - JOUR
AB - We prove the existence of scattering states for the defocusing cubic Gross-Pitaevskii (GP) hierarchy in ℝ3. Moreover, we show that an exponential energy growth condition commonly used in the well-posedness theory of the GP hierarchy is, in a specific sense, necessary. In fact, we prove that without the latter, there exist initial data for the focusing cubic GP hierarchy for which instantaneous blowup occurs.
AU - Chen, Thomas
AU - Hainzl, Christian
AU - Pavlović, Nataša
AU - Seiringer, Robert
ID - 2186
IS - 7
JF - Letters in Mathematical Physics
TI - On the well-posedness and scattering for the Gross-Pitaevskii hierarchy via quantum de Finetti
VL - 104
ER -
TY - JOUR
AB - Systems should not only be correct but also robust in the sense that they behave reasonably in unexpected situations. This article addresses synthesis of robust reactive systems from temporal specifications. Existing methods allow arbitrary behavior if assumptions in the specification are violated. To overcome this, we define two robustness notions, combine them, and show how to enforce them in synthesis. The first notion applies to safety properties: If safety assumptions are violated temporarily, we require that the system recovers to normal operation with as few errors as possible. The second notion requires that, if liveness assumptions are violated, as many guarantees as possible should be fulfilled nevertheless. We present a synthesis procedure achieving this for the important class of GR(1) specifications, and establish complexity bounds. We also present an implementation of a special case of robustness, and show experimental results.
AU - Bloem, Roderick
AU - Chatterjee, Krishnendu
AU - Greimel, Karin
AU - Henzinger, Thomas A
AU - Hofferek, Georg
AU - Jobstmann, Barbara
AU - Könighofer, Bettina
AU - Könighofer, Robert
ID - 2187
IS - 3-4
JF - Acta Informatica
TI - Synthesizing robust systems
VL - 51
ER -
TY - JOUR
AB - Although plant and animal cells use a similar core mechanism to deliver proteins to the plasma membrane, their different lifestyle, body organization and specific cell structures resulted in the acquisition of regulatory mechanisms that vary in the two kingdoms. In particular, cell polarity regulators do not seem to be conserved, because genes encoding key components are absent in plant genomes. In plants, the broad knowledge on polarity derives from the study of auxin transporters, the PIN-FORMED proteins, in the model plant Arabidopsis thaliana. In animals, much information is provided from the study of polarity in epithelial cells that exhibit basolateral and luminal apical polarities, separated by tight junctions. In this review, we summarize the similarities and differences of the polarization mechanisms between plants and animals and survey the main genetic approaches that have been used to characterize new genes involved in polarity establishment in plants, including the frequently used forward and reverse genetics screens as well as a novel chemical genetics approach that is expected to overcome the limitation of classical genetics methods.
AU - Kania, Urszula
AU - Fendrych, Matyas
AU - Friml, Jiřĺ
ID - 2188
IS - APRIL
JF - Open Biology
TI - Polar delivery in plants; commonalities and differences to animal epithelial cells
VL - 4
ER -
TY - CONF
AB - En apprentissage automatique, nous parlons d'adaptation de domaine lorsque les données de test (cibles) et d'apprentissage (sources) sont générées selon différentes distributions. Nous devons donc développer des algorithmes de classification capables de s'adapter à une nouvelle distribution, pour laquelle aucune information sur les étiquettes n'est disponible. Nous attaquons cette problématique sous l'angle de l'approche PAC-Bayésienne qui se focalise sur l'apprentissage de modèles définis comme des votes de majorité sur un ensemble de fonctions. Dans ce contexte, nous introduisons PV-MinCq une version adaptative de l'algorithme (non adaptatif) MinCq. PV-MinCq suit le principe suivant. Nous transférons les étiquettes sources aux points cibles proches pour ensuite appliquer MinCq sur l'échantillon cible ``auto-étiqueté'' (justifié par une borne théorique). Plus précisément, nous définissons un auto-étiquetage non itératif qui se focalise dans les régions où les distributions marginales source et cible sont les plus similaires. Dans un second temps, nous étudions l'influence de notre auto-étiquetage pour en déduire une procédure de validation des hyperparamètres. Finalement, notre approche montre des résultats empiriques prometteurs.
AU - Morvant, Emilie
ID - 2189
TI - Adaptation de domaine de vote de majorité par auto-étiquetage non itératif
VL - 1
ER -