TY - CONF AB - We study turn-based stochastic zero-sum games with lexicographic preferences over reachability and safety objectives. Stochastic games are standard models in control, verification, and synthesis of stochastic reactive systems that exhibit both randomness as well as angelic and demonic non-determinism. Lexicographic order allows to consider multiple objectives with a strict preference order over the satisfaction of the objectives. To the best of our knowledge, stochastic games with lexicographic objectives have not been studied before. We establish determinacy of such games and present strategy and computational complexity results. For strategy complexity, we show that lexicographically optimal strategies exist that are deterministic and memory is only required to remember the already satisfied and violated objectives. For a constant number of objectives, we show that the relevant decision problem is in NP∩coNP , matching the current known bound for single objectives; and in general the decision problem is PSPACE -hard and can be solved in NEXPTIME∩coNEXPTIME . We present an algorithm that computes the lexicographically optimal strategies via a reduction to computation of optimal strategies in a sequence of single-objectives games. We have implemented our algorithm and report experimental results on various case studies. AU - Chatterjee, Krishnendu AU - Katoen, Joost P AU - Weininger, Maximilian AU - Winkler, Tobias ID - 8272 SN - 03029743 T2 - International Conference on Computer Aided Verification TI - Stochastic games with lexicographic reachability-safety objectives VL - 12225 ER - TY - CHAP AB - The polymerization–depolymerization dynamics of cytoskeletal proteins play essential roles in the self-organization of cytoskeletal structures, in eukaryotic as well as prokaryotic cells. While advances in fluorescence microscopy and in vitro reconstitution experiments have helped to study the dynamic properties of these complex systems, methods that allow to collect and analyze large quantitative datasets of the underlying polymer dynamics are still missing. Here, we present a novel image analysis workflow to study polymerization dynamics of active filaments in a nonbiased, highly automated manner. Using treadmilling filaments of the bacterial tubulin FtsZ as an example, we demonstrate that our method is able to specifically detect, track and analyze growth and shrinkage of polymers, even in dense networks of filaments. We believe that this automated method can facilitate the analysis of a large variety of dynamic cytoskeletal systems, using standard time-lapse movies obtained from experiments in vitro as well as in the living cell. Moreover, we provide scripts implementing this method as supplementary material. AU - Dos Santos Caldas, Paulo R AU - Radler, Philipp AU - Sommer, Christoph M AU - Loose, Martin ED - Tran, Phong ID - 7572 SN - 0091679X T2 - Methods in Cell Biology TI - Computational analysis of filament polymerization dynamics in cytoskeletal networks VL - 158 ER - TY - JOUR AB - Most bacteria accomplish cell division with the help of a dynamic protein complex called the divisome, which spans the cell envelope in the plane of division. Assembly and activation of this machinery are coordinated by the tubulin-related GTPase FtsZ, which was found to form treadmilling filaments on supported bilayers in vitro1, as well as in live cells, in which filaments circle around the cell division site2,3. Treadmilling of FtsZ is thought to actively move proteins around the division septum, thereby distributing peptidoglycan synthesis and coordinating the inward growth of the septum to form the new poles of the daughter cells4. However, the molecular mechanisms underlying this function are largely unknown. Here, to study how FtsZ polymerization dynamics are coupled to downstream proteins, we reconstituted part of the bacterial cell division machinery using its purified components FtsZ, FtsA and truncated transmembrane proteins essential for cell division. We found that the membrane-bound cytosolic peptides of FtsN and FtsQ co-migrated with treadmilling FtsZ–FtsA filaments, but despite their directed collective behaviour, individual peptides showed random motion and transient confinement. Our work suggests that divisome proteins follow treadmilling FtsZ filaments by a diffusion-and-capture mechanism, which can give rise to a moving zone of signalling activity at the division site. AU - Baranova, Natalia S. AU - Radler, Philipp AU - Hernández-Rocamora, Víctor M. AU - Alfonso, Carlos AU - Lopez Pelegrin, Maria D AU - Rivas, Germán AU - Vollmer, Waldemar AU - Loose, Martin ID - 7387 JF - Nature Microbiology SN - 2058-5276 TI - Diffusion and capture permits dynamic coupling between treadmilling FtsZ filaments and cell division proteins VL - 5 ER - TY - JOUR AB - Fejes Tóth [3] studied approximations of smooth surfaces in three-space by piecewise flat triangular meshes with a given number of vertices on the surface that are optimal with respect to Hausdorff distance. He proves that this Hausdorff distance decreases inversely proportional with the number of vertices of the approximating mesh if the surface is convex. He also claims that this Hausdorff distance is inversely proportional to the square of the number of vertices for a specific non-convex surface, namely a one-sheeted hyperboloid of revolution bounded by two congruent circles. We refute this claim, and show that the asymptotic behavior of the Hausdorff distance is linear, that is the same as for convex surfaces. AU - Vegter, Gert AU - Wintraecken, Mathijs ID - 8163 IS - 2 JF - Studia Scientiarum Mathematicarum Hungarica SN - 0081-6906 TI - Refutation of a claim made by Fejes Tóth on the accuracy of surface meshes VL - 57 ER - TY - JOUR AB - We study relations between evidence theory and S-approximation spaces. Both theories have their roots in the analysis of Dempsterchr('39')s multivalued mappings and lower and upper probabilities, and have close relations to rough sets. We show that an S-approximation space, satisfying a monotonicity condition, can induce a natural belief structure which is a fundamental block in evidence theory. We also demonstrate that one can induce a natural belief structure on one set, given a belief structure on another set, if the two sets are related by a partial monotone S-approximation space. AU - Shakiba, A. AU - Goharshady, Amir Kafshdar AU - Hooshmandasl, M.R. AU - Alambardar Meybodi, M. ID - 8671 IS - 2 JF - Iranian Journal of Mathematical Sciences and Informatics SN - 1735-4463 TI - A note on belief structures and s-approximation spaces VL - 15 ER - TY - JOUR AB - The strong rate of convergence of the Euler-Maruyama scheme for nondegenerate SDEs with irregular drift coefficients is considered. In the case of α-Hölder drift in the recent literature the rate α/2 was proved in many related situations. By exploiting the regularising effect of the noise more efficiently, we show that the rate is in fact arbitrarily close to 1/2 for all α>0. The result extends to Dini continuous coefficients, while in d=1 also to all bounded measurable coefficients. AU - Dareiotis, Konstantinos AU - Gerencser, Mate ID - 6359 JF - Electronic Journal of Probability TI - On the regularisation of the noise for the Euler-Maruyama scheme with irregular drift VL - 25 ER - TY - THES AB - Deep neural networks have established a new standard for data-dependent feature extraction pipelines in the Computer Vision literature. Despite their remarkable performance in the standard supervised learning scenario, i.e. when models are trained with labeled data and tested on samples that follow a similar distribution, neural networks have been shown to struggle with more advanced generalization abilities, such as transferring knowledge across visually different domains, or generalizing to new unseen combinations of known concepts. In this thesis we argue that, in contrast to the usual black-box behavior of neural networks, leveraging more structured internal representations is a promising direction for tackling such problems. In particular, we focus on two forms of structure. First, we tackle modularity: We show that (i) compositional architectures are a natural tool for modeling reasoning tasks, in that they efficiently capture their combinatorial nature, which is key for generalizing beyond the compositions seen during training. We investigate how to to learn such models, both formally and experimentally, for the task of abstract visual reasoning. Then, we show that (ii) in some settings, modularity allows us to efficiently break down complex tasks into smaller, easier, modules, thereby improving computational efficiency; We study this behavior in the context of generative models for colorization, as well as for small objects detection. Secondly, we investigate the inherently layered structure of representations learned by neural networks, and analyze its role in the context of transfer learning and domain adaptation across visually dissimilar domains. AU - Royer, Amélie ID - 8390 SN - 2663-337X TI - Leveraging structure in Computer Vision tasks for flexible Deep Learning models ER - TY - CONF AB - Numerous methods have been proposed for probabilistic generative modelling of 3D objects. However, none of these is able to produce textured objects, which renders them of limited use for practical tasks. In this work, we present the first generative model of textured 3D meshes. Training such a model would traditionally require a large dataset of textured meshes, but unfortunately, existing datasets of meshes lack detailed textures. We instead propose a new training methodology that allows learning from collections of 2D images without any 3D information. To do so, we train our model to explain a distribution of images by modelling each image as a 3D foreground object placed in front of a 2D background. Thus, it learns to generate meshes that when rendered, produce images similar to those in its training set. A well-known problem when generating meshes with deep networks is the emergence of self-intersections, which are problematic for many use-cases. As a second contribution we therefore introduce a new generation process for 3D meshes that guarantees no self-intersections arise, based on the physical intuition that faces should push one another out of the way as they move. We conduct extensive experiments on our approach, reporting quantitative and qualitative results on both synthetic data and natural images. These show our method successfully learns to generate plausible and diverse textured 3D samples for five challenging object classes. AU - Henderson, Paul M AU - Tsiminaki, Vagia AU - Lampert, Christoph ID - 8186 T2 - Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition TI - Leveraging 2D data to learn textured 3D mesh generation ER - TY - JOUR AB - Earlier, we demonstrated that transcript levels of METAL TOLERANCE PROTEIN2 (MTP2) and of HEAVY METAL ATPase2 (HMA2) increase strongly in roots of Arabidopsis upon prolonged zinc (Zn) deficiency and respond to shoot physiological Zn status, and not to the local Zn status in roots. This provided evidence for shoot-to-root communication in the acclimation of plants to Zn deficiency. Zn-deficient soils limit both the yield and quality of agricultural crops and can result in clinically relevant nutritional Zn deficiency in human populations. Implementing Zn deficiency during cultivation of the model plant Arabidopsis thaliana on agar-solidified media is difficult because trace element contaminations are present in almost all commercially available agars. Here, we demonstrate root morphological acclimations to Zn deficiency on agar-solidified medium following the effective removal of contaminants. These advancements allow reproducible phenotyping toward understanding fundamental plant responses to deficiencies of Zn and other essential trace elements. AU - Sinclair, Scott A AU - Krämer, U. ID - 7416 IS - 1 JF - Plant Signaling & Behavior SN - 1559-2324 TI - Generation of effective zinc-deficient agar-solidified media allows identification of root morphology changes in response to zinc limitation VL - 15 ER - TY - JOUR AB - In order to provide a local description of a regular function in a small neighbourhood of a point x, it is sufficient by Taylor’s theorem to know the value of the function as well as all of its derivatives up to the required order at the point x itself. In other words, one could say that a regular function is locally modelled by the set of polynomials. The theory of regularity structures due to Hairer generalizes this observation and provides an abstract setup, which in the application to singular SPDE extends the set of polynomials by functionals constructed from, e.g., white noise. In this context, the notion of Taylor polynomials is lifted to the notion of so-called modelled distributions. The celebrated reconstruction theorem, which in turn was inspired by Gubinelli’s \textit {sewing lemma}, is of paramount importance for the theory. It enables one to reconstruct a modelled distribution as a true distribution on Rd which is locally approximated by this extended set of models or “monomials”. In the original work of Hairer, the error is measured by means of Hölder norms. This was then generalized to the whole scale of Besov spaces by Hairer and Labbé. It is the aim of this work to adapt the analytic part of the theory of regularity structures to the scale of Triebel–Lizorkin spaces. AU - Hensel, Sebastian AU - Rosati, Tommaso ID - 9196 IS - 3 JF - Studia Mathematica KW - General Mathematics SN - 0039-3223 TI - Modelled distributions of Triebel–Lizorkin type VL - 252 ER -