TY - GEN AB - Adaptive divergence and speciation may happen despite opposition by gene flow. Identifying the genomic basis underlying divergence with gene flow is a major task in evolutionary genomics. Most approaches (e.g. outlier scans) focus on genomic regions of high differentiation. However, not all genomic architectures potentially underlying divergence are expected to show extreme differentiation. Here, we develop an approach that combines hybrid zone analysis (i.e. focuses on spatial patterns of allele frequency change) with system-specific simulations to identify loci inconsistent with neutral evolution. We apply this to a genome-wide SNP set from an ideally-suited study organism, the intertidal snail Littorina saxatilis, which shows primary divergence between ecotypes associated with different shore habitats. We detect many SNPs with clinal patterns, most of which are consistent with neutrality. Among non-neutral SNPs, most are located within three large putative inversions differentiating ecotypes. Many non-neutral SNPs show relatively low levels of differentiation. We discuss potential reasons for this pattern, including loose linkage to selected variants, polygenic adaptation and a component of balancing selection within populations (which may be expected for inversions). Our work is in line with theory predicting a role for inversions in divergence, and emphasises that genomic regions contributing to divergence may not always be accessible with methods purely based on allele frequency differences. These conclusions call for approaches that take spatial patterns of allele frequency change into account in other systems. AU - Westram, Anja M AU - Rafajlović, Marina AU - Chaube, Pragya AU - Faria, Rui AU - Larsson, Tomas AU - Panova, Marina AU - Ravinet, Mark AU - Blomberg, Anders AU - Mehlig, Bernhard AU - Johannesson, Kerstin AU - Butlin, Roger ID - 9930 TI - Data from: Clines on the seashore: the genomic architecture underlying rapid divergence in the face of gene flow ER - TY - GEN AB - The evolution of assortative mating is a key part of the speciation process. Stronger assortment, or greater divergence in mating traits, between species pairs with overlapping ranges is commonly observed, but possible causes of this pattern of reproductive character displacement are difficult to distinguish. We use a multidisciplinary approach to provide a rare example where it is possible to distinguish among hypotheses concerning the evolution of reproductive character displacement. We build on an earlier comparative analysis that illustrated a strong pattern of greater divergence in penis form between pairs of sister species with overlapping ranges than between allopatric sister-species pairs, in a large clade of marine gastropods (Littorinidae). We investigate both assortative mating and divergence in male genitalia in one of the sister-species pairs, discriminating among three contrasting processes each of which can generate a pattern of reproductive character displacement: reinforcement, reproductive interference and the Templeton effect. We demonstrate reproductive character displacement in assortative mating, but not in genital form between this pair of sister species and use demographic models to distinguish among the different processes. Our results support a model with no gene flow since secondary contact and thus favour reproductive interference as the cause of reproductive character displacement for mate choice, rather than reinforcement. High gene flow within species argues against the Templeton effect. Secondary contact appears to have had little impact on genital divergence. AU - Hollander, Johan AU - Montaño-Rendón, Mauricio AU - Bianco, Giuseppe AU - Yang, Xi AU - Westram, Anja M AU - Duvaux, Ludovic AU - Reid, David G. AU - Butlin, Roger K. ID - 9929 TI - Data from: Are assortative mating and genital divergence driven by reinforcement? ER - TY - CONF AB - We introduce Intelligent Annotation Dialogs for bounding box annotation. We train an agent to automatically choose a sequence of actions for a human annotator to produce a bounding box in a minimal amount of time. Specifically, we consider two actions: box verification [34], where the annotator verifies a box generated by an object detector, and manual box drawing. We explore two kinds of agents, one based on predicting the probability that a box will be positively verified, and the other based on reinforcement learning. We demonstrate that (1) our agents are able to learn efficient annotation strategies in several scenarios, automatically adapting to the image difficulty, the desired quality of the boxes, and the detector strength; (2) in all scenarios the resulting annotation dialogs speed up annotation compared to manual box drawing alone and box verification alone, while also outperforming any fixed combination of verification and drawing in most scenarios; (3) in a realistic scenario where the detector is iteratively re-trained, our agents evolve a series of strategies that reflect the shifting trade-off between verification and drawing as the detector grows stronger. AU - Uijlings, Jasper AU - Konyushkova, Ksenia AU - Lampert, Christoph AU - Ferrari, Vittorio ID - 10882 SN - 9781538664209 T2 - 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition TI - Learning intelligent dialogs for bounding box annotation ER - TY - CONF AB - This paper studies the problem of distributed stochastic optimization in an adversarial setting where, out of m machines which allegedly compute stochastic gradients every iteration, an α-fraction are Byzantine, and may behave adversarially. Our main result is a variant of stochastic gradient descent (SGD) which finds ε-approximate minimizers of convex functions in T=O~(1/ε²m+α²/ε²) iterations. In contrast, traditional mini-batch SGD needs T=O(1/ε²m) iterations, but cannot tolerate Byzantine failures. Further, we provide a lower bound showing that, up to logarithmic factors, our algorithm is information-theoretically optimal both in terms of sample complexity and time complexity. AU - Alistarh, Dan-Adrian AU - Allen-Zhu, Zeyuan AU - Li, Jerry ID - 6558 T2 - Advances in Neural Information Processing Systems TI - Byzantine stochastic gradient descent VL - 2018 ER - TY - JOUR AB - The main result of this article is a generalization of the classical blossom algorithm for finding perfect matchings. Our algorithm can efficiently solve Boolean CSPs where each variable appears in exactly two constraints (we call it edge CSP) and all constraints are even Δ-matroid relations (represented by lists of tuples). As a consequence of this, we settle the complexity classification of planar Boolean CSPs started by Dvorak and Kupec. Using a reduction to even Δ-matroids, we then extend the tractability result to larger classes of Δ-matroids that we call efficiently coverable. It properly includes classes that were known to be tractable before, namely, co-independent, compact, local, linear, and binary, with the following caveat:We represent Δ-matroids by lists of tuples, while the last two use a representation by matrices. Since an n ×n matrix can represent exponentially many tuples, our tractability result is not strictly stronger than the known algorithm for linear and binary Δ-matroids. AU - Kazda, Alexandr AU - Kolmogorov, Vladimir AU - Rolinek, Michal ID - 6032 IS - 2 JF - ACM Transactions on Algorithms TI - Even delta-matroids and the complexity of planar boolean CSPs VL - 15 ER -