Local Map Research Articles

V-FCW: Vector-based forward collision warning algorithm for curved road conflicts using V2X networks

The implementation of advanced driver assistance systems (ADAS) has significantly impacted the prevention of traffic accidents, particularly through the forward collision warning (FCW) algorithm. Nevertheless, traffic conflicts on traffic routes remain a significant issue, since most FCW algorithms cannot accurately determine the distance between the host vehicle (HV) and remote vehicle (RV) on curved roads. Hence, this study proposes a vector-based FCW (V-FCW) algorithm to address the issue of false warnings on unconventional road sections. The V-FCW algorithm employs vector relationships to estimate the poses of HV and RV at the current and next moments, thereby effectively calculating the relative angles. Firstly, the HV and RV transmit their position vector, velocity vector, and heading angle in real time via the vehicle-to-vehicle (V2V) communication technique. Subsequently, the localization of lanes is conducted through the vehicle-to-infrastructure (V2I) communication technique, with the assistance of roadside unit (RSU)-based local maps. Finally, a V-FCW algorithm was implemented on the Simcenter Prescan simulation platform and a cellular vehicle-to-everything (C-V2X, i.e., the combination of V2V and V2I) communication platform. The simulation results demonstrate that the proposed V-FCW algorithm can accurately identify and warn dangerous vehicles on both straight and curved roads. Moreover, the experimental results obtained from the hardware-in-the-loop approach illustrate the efficacy of the proposed V-FCW algorithm in accurately forecasting four warning levels on both straight and curved roads. Consequently, this study yields a significant contribution to the field of vehicle-road cooperation in C-V2X-enable intelligent driving.

Effect of HRTFs and head motion on auditory-visual localization in real and virtual studio environments

When localizing sound sources in natural or virtual environments, multisensory integration that combines auditory and visual cues is often involved. This study aims to determine the effect of (non-)individual head-related transfer functions (HRTFs) on sound source localization in the presence of visual cues. Participants ($N=16$) were seated in a hemispherical loudspeaker arrangement, and their task was to indicate the perceived direction of pink-noise bursts. The conditions included localization of real loudspeakers and virtualized loudspeakers based on binaural rendering with (non-)individual HRTFs. The horizontal spacing of the visible loudspeakers was 30 degrees or larger and the vertical spacing was varied between 30 degrees and 15 degrees, to focus on vertical discrimination. The results of a static localization experiment (no head movements during playback) demonstrated an increased rate of vertical local confusions using non-individual binaural room impulse responses (BRIRs) of the KU100 dummy head compared to individual BRIRs. Dynamic auditory-visual localization with non-individual HRTFs suffered from vertical uncertainty and bias in the frontal area, where local confusion rates were consistent with the static experiment. In conclusion, the horizontal localization mapping of dynamic binaural rendering with non-individual HRTFs was comparable to real loudspeaker reproduction, while the vertical mapping was significantly impaired.

Stanisław Przybyszewski dzieciom Zapomniany, ostatni utwór pisarza

This article deals with ‘Nad Gopłem’ [At Lake Gopło], the least known and arguably the oddest story in the Przybyszewski oeuvre. The book – missing from the writer’s biblio-graphy and, as far as we know, not mentioned in his correspondence – was addressed exclusively to the young reader. What’s more, it was commissioned by the editors of a Polish language teaching textbook (published by Ossolineum in 1928) from a writer with a reputation of a debaucher and corruptor of the young. The article points out that Przybyszewski’s text is an inspirational tale about the beauty spots of Kujawy and a story about the education of an artist. Showing that Poland is born in a personal, intimate and creative experience and beauty lies in the eye of the beholder, he pins down the sources of his writing onto a local map and argues that the artist’s ties with his homeland are unbreakable. The loose string d of reminiscences creates a warm, atmospheric portrait of the landscapes around Lake Gopło and, thanks to unobtrusive emotional touches, succeeds in eliciting a similar affection in the reader. In this way the story meets per-fectly the contemporary postulate of anchoring the educational goals in reading materi-als “based on the experiences of a child” and reveals an unknown facet of Przybyszew-ski’s persona you would not expect to find in a celebrity engulfed in scandal. He knows how to strike the right tone and find the right format to “talk” to children and gently touch their emotions.

Innovative Method for Reliable Measurement of PEM Water Electrolyzer Component Resistances.

Understanding the sheet resistance of porous electrodes is essential for improving the performance of polymer electrolyte membrane (PEM) water electrolyzers and related technologies. Despite its importance, existing methods often fail to provide reliable and comprehensive data, especially for porous materials with complex morphologies and non-uniform thicknesses. This study introduces a robust and straightforward method for determining the sheet resistance of porous electrodes using a novel probe concept based on industrial printed circuit board (PCB) technology. This probe measures resistance across ten distances, ranging from 250 µm to 2500µm, enabling local mapping of resistance. The study focuses on the sheet resistance of key components in PEM water electrolyzers, including the gas diffusion layer (GDL), porous transport layer (PTL), and catalyst layers deposited on a membrane. Additionally, an image-processing-based method is presented to obtain the thickness distribution of the studied catalyst layers, facilitating a detailed analysis of the electrical in-plane resistivity with thickness variations. Overall, this methodology has the potential to expedite material integration and bridge the gap between electrode engineering and single-cell testing, thereby advancing the development of PEM water electrolyzers.

FL-W3S: Cross-domain federated learning for weakly supervised semantic segmentation of white blood cells.

Segmentation models for clinical data experience severe performance degradation when trained on a single client from one domain and distributed to other clients from different domain. Federated Learning (FL) provides a solution by enabling multi-party collaborative learning without compromising the confidentiality of clients' private data. In this paper, we propose a cross-domain FL method for Weakly Supervised Semantic Segmentation (FL-W3S) of white blood cells in microscopic images. We perform model training on multiple clients with different data distributions to obtain a global aggregated model using only image-level class labels for semantic segmentation of white blood cells. A multi-class token transformer model learns the relationship between patch tokens and class tokens during collaborative learning and generates class-specific localization maps for mask predictions. To rectify the localization maps, we use patch-level pairwise affinity obtained from patch-to-patch transformer attention. We evaluate performance of the proposed semantic segmentation method on two different datasets of white blood cells from different domains. Our experimental results show that for two datasets, there is 2.56% and 1.39% increase in performance of the proposed method over existing state-of-the-art methods. The combination of federated learning for collaborative model training while preserving data privacy, alongside white blood cell segmentation techniques for precise cell identification, enhances diagnostic accuracy and personalized treatment strategies in clinical applications, particularly in hematology and pathology. More specifically, it involves isolating white blood cell from blood smear for further analysis such as automated blood cell counting, morphological analysis, cell classification, disease diagnosis and monitoring.

Swin-transformer for weak feature matching

Feature matching in computer vision is crucial but challenging in weakly textured scenes due to the lack of pattern repetition. We introduce the SwinMatcher feature matching method, aimed at addressing the issues of low matching quantity and poor matching precision in weakly textured scenes. Given the inherently significant local characteristics of image features, we employ a local self-attention mechanism to learn from weakly textured areas, maximally preserving the features of weak textures. To address the issue of incorrect matches in scenes with repetitive patterns, we use a cross-attention and positional encoding mechanism to learn the correct matches of repetitive patterns in two scenes, achieving higher matching precision. We also introduce a matching optimization algorithm that calculates the spatial expected coordinates of local two-dimensional heat maps of correspondences to obtain the final sub-pixel level matches. Experiments indicate that, under identical training conditions, the SwinMatcher outperforms other standard methods in pose estimation, homography estimation, and visual localization. It exhibits strong robustness and superior matching in weakly textured areas, offering a new research direction for feature matching in weakly textured images.

Study on automated guided vehicle collision avoidance mechanism with external computer vision

In industrial automation, safely operating Automated Guided Vehicles (AGVs) in dynamic environments is challenging. Traditional collision avoidance methods, which rely on technologies like LiDAR (Light Detection and Ranging), have limitations in detection range, accuracy, and cost. This study introduces a new method called the Collision Avoidance System with External Computer Vision (CAECV). CAECV uses two cameras which are equipped on the wall in the factory to detect obstacles, and AGV don’t need to equip any obstacle detection sensors. The detection algorithm, combining YOLOv8 with an attention mechanism, excels at identifying obstacles. The system then converts image coordinates to global coordinates using inverse perspective mapping for precise obstacle localization. Experimental comparisons with traditional TIM laser scanning radar show that CAECV outperforms in detection range, response time, accuracy, and cost. CAECV offers a broader detection range, quicker response, and lower error rates. As more AGVs are deployed, CAECV becomes significantly more cost-effective than LiDAR-based systems. Moreover, CAECV can be integrated with LiDAR systems to enhance AGV obstacle avoidance. CAECV represents a scalable, accurate, and economical advancement in AGV collision avoidance technology for modern industrial use.

Cyclic Peak Extraction from a Spatial Likelihood Map for Multi-Array Multi-Target Bearing-Only Localization

In the context of multi-array multi-target bearing-only localization, due to the existence of direction-finding errors, the crossing results of bearing lines cannot accurately determine correspondence with targets. Under conditions that clutter interference and missing of detection in direction-finding, the traditional method will produce false alarm targets and miss some targets. To address this issue, this paper draws on the idea of a spatial likelihood map which calculates the likelihood of target presence at each grid point within the observation area by partitioning the observation area into grids and utilizing bearing data from each array, yielding the distribution of targets in the observation area. Then, a multi-target cyclic peak extraction algorithm based on a statistical dual-threshold is proposed, which eliminates false peaks by cyclic extraction of target positions, so as to reduce false targets. Simulation results demonstrate that the spatial likelihood mapping-based localization exhibits good performance. Furthermore, when the multi-target cyclic peak extraction algorithm based on statistical dual-thresholds is applied, it outperforms direct target extraction from the spatial likelihood map, showcasing enhanced multi-target localization capabilities. Moreover, compared to the position non-linear least squares multi-target localization method, the proposed method has lower optimal sub-pattern assignment distance and lower localization error under the condition of interference and missing detection.

Ultrastructural Localization of Glutamate Delta Receptor 1 in the Rodent and Primate Lateral Habenula.

Glutamate delta receptor 1 (GluD1) is a unique synaptogenic molecule expressed at excitatory and inhibitory synapses. The lateral habenula (LHb), a subcortical structure that regulates negative reward prediction error and major monoaminergic systems, is enriched in GluD1. LHb dysfunction has been implicated in psychiatric disorders such as depression and schizophrenia, both of which are associated with GRID1, the gene that encodes GluD1. Thus, disruption in GluD1 synaptic signaling may contribute to LHb dysfunction and the pathophysiology of LHb-associated disorders. Despite its strong cellular expression, little is known about the subsynaptic and subcellular localization of GluD1 in LHb neurons. Given that GluD1 is involved in the development and/or regulation of glutamatergic and GABAergic synapses in various brain regions, a detailed map of GluD1 synaptic localization is essential to elucidate its role in the LHb. To address this issue, we used immunoelectron microscopy methods in rodents and monkeys. In both species, GluD1 immunoreactivity was primarily expressed in dendritic profiles, with lower expression in somata, spines, and glial elements. Pre- and post-embedding immunogold experiments revealed strong GluD1 expression in the core of symmetric GABAergic synapses. Albeit less frequent, GluD1 was also found at the edges (i.e., perisynaptic) of asymmetric, putative glutamatergic synapses. Through the combination of anterograde tracing with immunogold labeling in rats, we showed that axon terminals from the entopeduncular nucleus and the lateral hypothalamus express postsynaptic GluD1 immunolabeling in the LHb. Our findings suggest that GluD1 may play a critical role in modulating GABAergic transmission in the rodent and primate LHb.

Thin free-standing liquid films manipulation: device design to turn on/off gravity in flow regimes for thickness map control and for material structuring.

The manipulation and control of free-standing liquid film drainage dynamics is of paramount importance in many technological fields and related products, ranging from liquid lenses to liquid foams and 2D structures. In this context, we theoretically design and introduce a device where we can reversibly drive flow regime switch between viscous-capillary and viscous-gravity in a thin free-standing liquid film by altering its shape, allowing us to manipulate and stabilize the film thickness over time. The device, which mainly consists of a syringe pump, a pressure transducer, and a 3D-printed cylinder, is coupled with a digital holography setup to measure, in real time, the evolution of the local film thickness map, revealing characteristic features of viscous-capillary and viscous-gravity driven drainage regimes. By using polyvinyl alcohol/water concentrated solutions, we are also able to produce viscoelastic membranes after manipulation and water evaporation, which presents manipulation history-dependent geometrical properties. Furthermore, using a system composed of carboxymethyl cellulose, water, and rod-like zinc oxide nanoparticles, we show a clear effect of film manipulation on particle rearrangement. We believe this device could represent a starting point for the development of a useful and practical tool to study thin liquid film dynamics and to produce novel (patterned) 2D structures for the numerous scientific and technical fields where they are of use.

Assessing Noise Impact on DeepOrientation - A Convolutional Neural Network for Local Fringe Orientation Map Estimation

This paper discusses noise robustness of DeepOrientation, a convolutional neural network developed for fast and accurate local fringe orientation map estimation, enhancing full-field optical measurement techniques such as interferometry and holographic microscopy. The use of neural networks to determine the final result of the optical measurement may raise legitimate metrological concerns and therefore we still recommend the use of fully mathematically sound solutions for both fringe pattern prefiltration and phase retrieval. DeepOrientation does not replace mathematically rigorous algorithms but supports them providing fringe orientation map vital for 2D Hilbert transform phase demodulation, requiring prefiltered fringe data for optimal performance. We analyze DeepOrientation sensitivity to prefiltration noise-related accuracy using simulated data and validate results with experimentally recorded fringe patterns. Full Text: PDF References G. Popescu, Quantitative Phase Imaging of Cells and Tissues, (McGraw Hill Professional, 2011). DirectLink D. Malacara, M. Servin, Z. Malacara, Interferogram Analysis for Optical Testing (Marcel Dekker, 1998). DirectLink M. Takeda, H. Ina, S. Kobayashi, "Fourier-transform method of fringe-pattern analysis for computer-based topography and interferometry", J. Opt. Soc. Am. 72(1), 156 (1982). CrossRef Q. Kemao, "Windowed Fourier transform for fringe pattern analysis", Appl. Opt. 43(13), 2695 (2004). CrossRef K.G. Larkin, D.J. Bone, M.A. Oldfield, "Natural demodulation of two-dimensional fringe patterns. I. General background of the spiral phase quadrature transform", J. Opt. Soc. Am. A 18(8), 1862 (2001). CrossRef M. Trusiak, M. Cywinska, V. Mico, J.-A. Picazo-Bueno, C. Zuo, P. Zdankowski, K. Patorski, "Variational Hilbert Quantitative Phase Imaging", Sci Rep 10, 13955 (2020). CrossRef B. Jahne, Practical Handbook on Image Processing for Scientific Applications (CRC, 1997). CrossRef F. Zhang, W. Liu, J. Wang, Y. Zhu, and L. Xia, "Anisotropic partial differential equation noise-reduction algorithm based on fringe feature for ESPI", Opt. Commun. 282(12), 2318 (2009). CrossRef J. Villa, J. A. Quiroga, and I. De la Rosa, "Regularized quadratic cost function for oriented fringe-pattern filtering", Opt. Lett. 34(11), 1741 (2009). CrossRef Q. Yu, X. Liu, and X. Sun, "Generalized spin filtering and an improved derivative-sign binary image method for the extraction of fringe skeletons", Appl. Opt. 37(20), 4504 (1998). CrossRef H. Wang, Q. Kemao, "Frequency guided methods for demodulation of a single fringe pattern", Opt. Express 17(17), 15118 (2009). CrossRef J.L. Marroquin, R. Rodriguez-Vera, M. Servin, "Local phase from local orientation by solution of a sequence of linear systems", J. Opt. Soc. Am. A 15(6), 1536 (1998). CrossRef M. Cywińska, M. Rogalski, F. Brzeski, K. Patorski, M. Trusiak, "DeepOrientation: convolutional neural network for fringe pattern orientation map estimation", Opt. Express 30(23), 42283 (2022). CrossRef M. Cywińska, M. Trusiak, K. Patorski, "Automatized fringe pattern preprocessing using unsupervised variational image decomposition", Opt. Express 27(16), 22542 (2019). CrossRef X. Yang, Q. Yu, S. Fu, "A combined method for obtaining fringe orientations of ESPI", Opt. Commun. 273(1), 60 (2007). CrossRef

Robust Image Forgery Detection and Localization Framework using Vision Transformers (ViTs)

Image forgery detection has become increasingly critical with the proliferation of image editing tools capable of generating realistic forgeries. Traditional deep learning approaches, such as convolutional neural networks (CNNs), often struggle with capturing global dependencies and subtle inconsistencies across larger image contexts. To address these challenges, this paper proposes a novel Vision Transformer(ViT)- based framework for robust image forgery detection and localization. Leveraging the self-attention mechanism of transformers, our approach effectively models long-range dependencies and detects even subtle tampered regions with high precision. The proposed framework processes images as patch embeddings, extracting both local and global features, and outputs a detailed forgery map for accurate localization. We evaluate our method on multiple benchmark datasets containing diverse forgery types, including splicing, cloning, and inpainting. Experimental results demonstrate that the Vit based model outperforms state-of-the-art CNN and GAN-based methods, achieving superior accuracy, precision, and recall. Additionally, qualitative analyses highlight its capability to localize forgeries in complex scenarios. The results underscore the potential of Vision Transformers as a powerful tool for advancing the field of image forgery detection.

Machine learning-driven wind energy mapping enhanced by natural neighbor interpolation

In the present work, a prediction on the wind energy potential in Semarang City (Central Java Province, Indonesia) has been performed by leveraging a novel combination of machine learning and natural neighbor interpolation (NNI) methodology. This integrated approach uniquely combines the predictive power of machine learning to estimate wind speeds based on historical and spatial data, with the spatial mapping capabilities of NNI, which provides a more accurate and seamless visualization of wind speed distribution. This combination addresses challenges of data sparsity and variability, offering a more reliable and localized mapping approach than traditional methods. Additionally, air density is considered to calculate energy density, enabling a comprehensive evaluation of wind energy potential. The results show an average monthly wind speed of 5.23 m/s, ranging from 3.38 m/s to 7.39 m/s. Wind speeds between 7 m/s and 10 m/s are predicted to occur for up to 10 months annually, with an estimated energy density of 102.7 W/m². These findings underscore the feasibility of small-scale wind power generation in the study area and provide actionable insights for advancing renewable energy policies and implementations at the local level.

Local Gradients of Functional Connectivity Enable Precise Fingerprinting of Infant Brains During Dynamic Development.

Brain functional connectivity patterns exhibit distinctive, individualized characteristics capable of distinguishing one individual from others, like fingerprint. Accurate and reliable depiction of individualized functional connectivity patterns during infancy is crucial for advancing our understanding of individual uniqueness and variability of the intrinsic functional architecture during dynamic early brain development, as well as its role in neurodevelopmental disorders. However, the highly dynamic and rapidly developing nature of the infant brain presents significant challenges in capturing robust and stable functional fingerprint, resulting in low accuracy in individual identification over ages during infancy using functional connectivity. Conventional methods rely on brain parcellations for computing inter-regional functional connections, which are sensitive to the chosen parcellation scheme and completely ignore important fine-grained, spatially detailed patterns in functional connectivity that encodes developmentally-invariant, subject-specific features critical for functional fingerprinting. To solve these issues, for the first time, we propose a novel method to leverage the high-resolution, vertex-level local gradient map of functional connectivity from resting-state functional MRI, which captures sharp changes and subject-specific rich information of functional connectivity patterns, to explore infant functional fingerprint. Leveraging a longitudinal dataset comprising 591 high-resolution resting-state functional MRI scans from 103 infants, our method demonstrates superior performance in infant individual identification across ages. Our method has unprecedentedly achieved 99% individual identification rates across three age-varied sub-datasets, with consistent and robust identification rates across different phase encoding directions, significantly outperforming atlas-based approaches with only around 70% accuracy. Further vertex-wise uniqueness and differential power analyses highlighted the discriminative identifiability of higher-order functional networks. Additionally, the local gradient-based functional fingerprints demonstrated reliable predictive capabilities for cognitive performance during infancy. These findings suggest the existence of unique individualized functional fingerprints during infancy and underscore the potential of local gradients of functional connectivity in capturing neurobiologically meaningful and fine-grained features of individualized characteristics for advancing normal and abnormal early brain development.

Adaptive Ultra-Wideband/Pedestrian Dead Reckoning Localization Algorithm Based on Maximum Point-by-Point Distance

Positioning using ultra-wideband (UWB) signals can be used to achieve centimeter-level indoor positioning. UWB has been widely used in indoor localization, vehicle networking, industrial IoT, etc. However, due to non-line-of-sight (NLOS) and multipath interference problems, UWB cannot provide adequate position information, which affects the final positioning accuracy. This paper proposes an adaptive UWB/PDR localization algorithm based on the maximum point-by-point distance to solve the problems of poor UWB performance and the error accumulation of the pedestrian dead reckoning (PDR) algorithm in NLOS scenarios that is used to enhance the robustness and accuracy of indoor positioning. Specifically, firstly, the cumulative distribution function (CDF) map of localization under normal conditions is obtained through offline pretraining and then compared with the CDF obtained when pedestrians are moving on the line. Then, the maximum point-by-point distance algorithm is used to identify the abnormal base stations. Then, the standard base stations are filtered out for localization. To further improve the localization accuracy, this paper proposes a UWB/PDR algorithm based on an improved adaptive extended Kalman filtering (EKF), which dynamically adjusts the position information through the adaptive factor, eliminates the influence of significant errors on the current position information and realizes multi-sensor fusion positioning. The realization results show that the algorithm in this paper has a solid ability to identify abnormal base stations and that the adaptive extended Kalman filtering (AEKF) algorithm is improved by 81.27%, 58.50%, 29.76%, and 18.06% compared to the PDR, UWB, EKF, and AEKF algorithms, respectively.

A semantic notation for comparing global high-resolution coastal flooding studies

IntroductionGlobal coastal flooding maps are now achieving a level of detail suitable for local applications. The resolution of these maps, derived from widely available open data sources, is approaching that of local flooding maps (0.5–100 m), increasing the need for a standardized approach to evaluate underlying assumptions and indicators for local applications.MethodsThis study introduces the Waterlevel, Elevation, Protection, Flood, Impact, Future (WEPFIF) notation, a structured notation for documenting and comparing key methodological choices and data variations across global coastal flooding studies. This approach enhances the understanding and explanation of the fitness-for- purpose of flood maps. This notation builds on commonly used methodological choices, dataset variations, and model approaches in global flooding risk research. Analysis of these workflows identifies common elements and highlights the need for a more structured reporting approach to improve comparability.ResultsApplying the WEPFIF notation to a case study in the Netherlands reveals significant variations in flood risk assessments originating from differences in Digital Elevation Model (DEM) and water level selection, and inclusion of protective infrastructure.DiscussionWEPFIF, by annotating these methodological variations, enables more informed comparisons between local and global flood studies. This allows researchers and practitioners to select appropriate data and models, based on their specific research objectives. The study proposes tailored approaches for three common types of flood studies: raising concern, optimizing flood protection investments, and representing the state of coastal risk.

Multi‐indice B$B$‐series

AbstractWe propose a novel way to study numerical methods for ordinary differential equations in one dimension via the notion of multi‐indice. The main idea is to replace rooted trees in Butcher's ‐series by multi‐indices. The latter were introduced recently in the context of describing solutions of singular stochastic partial differential equations. The combinatorial shift away from rooted trees allows for a compressed description of numerical schemes. Furthermore, such multi‐indices ‐series uniquely characterize the Taylor expansion of 1‐dimensional local and affine equivariant maps.

Deep Learning and Visual SLAM for Autonomous Navigation of UAVs: Status, Challenges, and Future Perspectives

This paper discusses the application and integration of deep learning and visual SLAM technology in UAV autonomous navigation. With the wide application of UAVs in the fields of transportation, agriculture, military and environmental monitoring, improving the autonomous navigation capabilities of UAVs has become an important demand. This paper first analyses the advantages and disadvantages of deep learning and visual SLAM. Then, this paper emphasizes the core role of deep learning in feature extraction, target recognition and path planning. The key role of visual SLAM in real-time localization and environment mapping is highlighted as well. This paper discusses the combined application of these two technologies and demonstrates how to enhance the stability and accuracy of visual SLAM systems through deep learning in complex dynamic environments. Although the current technology still faces high computational requirements and real-time challenges, this paper proposes future research direction to promote the continued development of UAV autonomous navigation technology.

Construction of a physical map for Aegilops geniculata chromosome 7Mg and localization of its novel purple coleoptile gene.

A physical map of Aegilops geniculata chromosome 7Mg was constructed, and a novel purple coleoptile gene was localized at 7MgS bin FL 0.60-0.65 by development of wheat-Ae. geniculata structural aberrations. The development of wheat-wild relative chromosomal structure aberrations not only provides novel germplasm resources for wheat improvement, but also aids in mapping desirable genes to specific chromosomal regions. Aegilops geniculata (2n = 4x = 28, UgUgMgMg), a wild relative of common wheat, possesses many favorable genes. In this study, Ae. geniculata chromosome 7Mg was identified as harboring a purple coleoptile gene by phenotypic evaluation of Chinese Spring (CS)-Ae. geniculata addition and substitution lines. To construct a physical map of chromosome 7Mg and localize the purple coleoptile gene, 59 molecular markers specific to 7Mg were developed, and 43 wheat-Ae. geniculata 7Mg chromosome structure aberrations were generated based on chromosome centromeric breakage-fusion and ph1b-induced homoeologous recombination. Segment sizes and breakpoint positions of each 7Mg structure aberration were further characterized using in situ hybridization and molecular marker analysis. Consequently, a physical map of chromosome 7Mg was constructed with 59 molecular markers, comprising six bins with 28 markers on 7MgS and six bins with 31 markers on 7MgL, and the purple coleoptile gene was mapped to an interval of FL 0.60-0.65 on 7MgS. The newly developed wheat-Ae. geniculata 7Mg structural aberrations and the physical map of 7Mg will facilitate the transfer and utilization of desirable genes from 7Mg in the future.

Bayesian nonparametric mixture of experts for inverse problems

Large classes of problems can be formulated as inverse problems, where the goal is to find parameter values that best explain some observed measures. The relationship between parameters and observations is typically highly non-linear, with relatively high dimensional observations and correlated multidimensional parameters. To deal with these constraints via inverse regression strategies, we consider the Gaussian Local Linear Mapping (GLLiM) model, a special instance of mixture of expert models. We propose a general scheme to design a Bayesian nonparametric GLLiM model to avoid any commitment to an arbitrary number of experts. A tractable estimation algorithm is designed using variational Bayesian expectation-maximisation. We establish posterior consistency for the number of mixture components after the merge-truncate-merge algorithm post-processing. Illustrations on simulated data show good results in terms of recovering the true number of experts and the regression function.