Advanced search in
Research products
arrow_drop_down
Searching FieldsTerms
Any field
arrow_drop_down
includes
arrow_drop_down
Include:
2,700 Research products, page 1 of 270

  • Publications
  • Research data
  • Research software
  • 2013-2022
  • Open Access
  • Preprint
  • AE

10
arrow_drop_down
Date (most recent)
arrow_drop_down
  • Open Access
    Authors: 
    Nadia Figueroa; Haiwei Dong; Abdulmotaleb El Saddik;
    Publisher: Association for Computing Machinery (ACM)
    Country: Switzerland

    We propose a 6D RGB-D odometry approach that finds the relative camera pose between consecutive RGB-D frames by keypoint extraction and feature matching both on the RGB and depth image planes. Furthermore, we feed the estimated pose to the highly accurate KinectFusion algorithm, which uses a fast ICP (Iterative Closest Point) to fine-tune the frame-to-frame relative pose and fuse the depth data into a global implicit surface. We evaluate our method on a publicly available RGB-D SLAM benchmark dataset by Sturm et al. The experimental results show that our proposed reconstruction method solely based on visual odometry and KinectFusion outperforms the state-of-the-art RGB-D SLAM system accuracy. Moreover, our algorithm outputs a ready-to-use polygon mesh (highly suitable for creating 3D virtual worlds) without any postprocessing steps.

  • Open Access
    Authors: 
    Christian Boudreault; Hichem Eleuch; Michael Hilke; Richard MacKenzie;
    Publisher: American Physical Society (APS)
    Project: NSERC

    One of the most challenging problems for the realization of a scalable quantum computer is to design a physical device that keeps the error rate for each quantum processing operation low. These errors can originate from the accuracy of quantum manipulation, such as the sweeping of a gate voltage in solid state qubits or the duration of a laser pulse in optical schemes. Errors also result from decoherence, which is often regarded as more crucial in the sense that it is inherent to the quantum system, being fundamentally a consequence of the coupling to the external environment. Grouping small collections of qubits into clusters with symmetries can protect parts of the calculation from decoherence. We use 4-level cores with a straightforward generalization of discrete rotational symmetry, omega-rotation invariance, to encode pairs of coupled qubits and universal 2-qubit logical gates. We include quantum errors as a main source of decoherence, and show that symmetry makes logical operations particularly resilient to untimely anisotropic qubit rotations. We propose a scalable scheme for universal quantum computation where cores play the role of quantum-computational transistors, quansistors. Initialization and readout are achieved by coupling to leads. The external leads are explicitly considered and are assumed to be the other main source of decoherence. We show that quansistors can be dynamically decoupled from the leads by tuning their internal parameters, giving them the versatility required to act as controllable quantum memory units. With this dynamical decoupling, logical operations within quansistors are also symmetry-protected from unbiased noise in their parameters. We identify technologies that could implement omega-rotation invariance. Many of our results can be generalized to higher-level omega-rotation-invariant systems, or adapted to clusters with other symmetries. Comment: 23 pages, 19 figures

  • Open Access
    Authors: 
    Montasir Qasymeh; Hichem Eleuch;
    Publisher: Springer Science and Business Media LLC

    Quantum microwave transmission is key to realizing modular superconducting quantum computers and distributed quantum networks. A large number of incoherent photons are thermally generated within the microwave frequency spectrum. The closeness of the transmitted quantum state to the source-generated quantum state at the input of the transmission link (measured by the transmission fidelity) degrades due to the presence of the incoherent photons. Hence, high-fidelity quantum microwave transmission has long been considered to be infeasible without refrigeration [3,4]. In this study, we propose a novel method for high-fidelity quantum microwave transmission using a room-temperature lossy waveguide. The proposed scheme consists of connecting two cryogenic nodes (i.e., a transmitter and a receiver) by the room-temperature lossy microwave waveguide. First, cryogenic preamplification is implemented prior to transmission. Second, at the receiver side, a cryogenic loop antenna is placed inside the output port of the waveguide and coupled to an LC harmonic oscillator located outside the waveguide. The loop antenna converts quantum microwave fields (which contain both signal and noise photons) to a quantum voltage across the coupled LC harmonic oscillator. The loop antenna detector at the receiver is designed to extensively suppress the induced photons across the LC oscillator. The signal transmittance is maintained intact by providing significant preamplification gain. Our calculations show that high-fidelity quantum transmission (i.e., more than 95%) is realized based on the proposed scheme for transmission distances reaching 100 m. Comment: 10pages; 6 figures

  • Open Access
    Authors: 
    Filippo Macchi; Eric Edsinger; Kirsten C. Sadler;
    Publisher: Cold Spring Harbor Laboratory

    Abstract Background Epigenetic regulatory mechanisms are divergent across the animal kingdom, yet these mechanisms are not well studied in non-model organisms. Unique features of cephalopods make them attractive for investigating behavioral, sensory, developmental, and regenerative processes, and recent studies have elucidated novel features of genome organization and gene and transposon regulation in these animals. However, it is not known how epigenetics regulates these interesting cephalopod features. We combined bioinformatic and molecular analysis of Octopus bimaculoides to investigate the presence and pattern of DNA methylation and examined the presence of DNA methylation and 3 histone post-translational modifications across tissues of three cephalopod species. Results We report a dynamic expression profile of the genes encoding conserved epigenetic regulators, including DNA methylation maintenance factors in octopus tissues. Levels of 5-methyl-cytosine in multiple tissues of octopus, squid, and bobtail squid were lower compared to vertebrates. Whole genome bisulfite sequencing of two regions of the brain and reduced representation bisulfite sequencing from a hatchling of O. bimaculoides revealed that less than 10% of CpGs are methylated in all samples, with a distinct pattern of 5-methyl-cytosine genome distribution characterized by enrichment in the bodies of a subset of 14,000 genes and absence from transposons. Hypermethylated genes have distinct functions and, strikingly, many showed similar expression levels across tissues while hypomethylated genes were silenced or expressed at low levels. Histone marks H3K27me3, H3K9me3, and H3K4me3 were detected at different levels across tissues of all species. Conclusions Our results show that the DNA methylation and histone modification epigenetic machinery is conserved in cephalopods, and that, in octopus, 5-methyl-cytosine does not decorate transposable elements, but is enriched on the gene bodies of highly expressed genes and could cooperate with the histone code to regulate tissue-specific gene expression.

  • Open Access English
    Authors: 
    Taushif Khan; Mahbubur Rahman; Ishfaq Ahmed; F. Al Ali; Puthen V. Jithesh; Nico Marr;
    Publisher: Frontiers Media SA

    Allelic diversity of human leukocyte antigen (HLA) class II genes may help maintain humoral immunity against infectious diseases. In this study, we investigated germline genetic variation in classical HLA class II genes and employed a systematic, unbiased approach to explore the relative contribution of this genetic variation in the antibody repertoire to various common pathogens. We leveraged a well-defined cohort of 800 adults representing the general Arab population in which genetic material is shared because of the high frequency of consanguineous unions. By applying a high-throughput method for large-scale antibody profiling to this well-defined cohort, we were able to dissect the overall effect of zygosity for classical HLA class II genes, as well as the effects associated with specific HLA class II alleles, haplotypes and genotypes, on the antimicrobial antibody repertoire breadth and antibody specificity with unprecedented resolution. Our population genetic studies revealed that zygosity of the classical HLA class II genes is a strong predictor of antibody responses to common human pathogens, suggesting that classical HLA class II gene heterozygosity confers a selective advantage. Moreover, we demonstrated that multiple HLA class II alleles can have additive effects on the antibody repertoire to common pathogens. We also identified associations of HLA-DRB1 genotypes with specific antigens. Our findings suggest that HLA class II gene polymorphisms confer specific humoral immunity against common pathogens, which may have contributed to the genetic diversity of HLA class II loci during hominine evolution.

  • Open Access
    Authors: 
    Tarun Dutta; Adrián Pérez-Salinas; Jasper Phua Sing Cheng; José Ignacio Latorre; Manas Mukherjee;
    Publisher: American Physical Society (APS)

    Quantum computers can provide solutions to classically intractable problems under specific and adequate conditions. However, current devices have only limited computational resources, and an effort is made to develop useful quantum algorithms under these circumstances. This work experimentally demonstrates that a single-qubit device can host a universal classifier. The quantum processor used in this work is based on ion traps, providing highly accurate control on small systems. The algorithm chosen is the re-uploading scheme, which can address general learning tasks. Ion traps suit the needs of accurate control required by re-uploading. In the experiment here presented, a set of non-trivial classification tasks are successfully carried. The training procedure is performed in two steps combining simulation and experiment. Final results are benchmarked against exact simulations of the same method and also classical algorithms, showing a competitive performance of the ion-trap quantum classifier. This work constitutes the first experimental implementation of a classification algorithm based on the re-uploading scheme. 13 pages, 11 figures, and 1 table

  • Publication . Preprint . Article . 2022
    Open Access
    Authors: 
    Yan, Yichao; Li, Jinpeng; Liao, Shengcai; Qin, Jie; Ni, Bingbing; Yang, Xiaokang; Shao, Ling;
    Publisher: Association for the Advancement of Artificial Intelligence (AAAI)

    Person search has recently emerged as a challenging task that jointly addresses pedestrian detection and person re-identification. Existing approaches follow a fully supervised setting where both bounding box and identity annotations are available. However, annotating identities is labor-intensive, limiting the practicability and scalability of current frameworks. This paper inventively considers weakly supervised person search with only bounding box annotations. We proposed to address this novel task by investigating three levels of context clues (i.e., detection, memory and scene) in unconstrained natural images. The first two are employed to promote local and global discriminative capabilities, while the latter enhances clustering accuracy. Despite its simple design, our CGPS achieves 80.0% in mAP on CUHK-SYSU, boosting the baseline model by 8.8%. Surprisingly, it even achieves comparable performance with several supervised person search models. Our code is available at https://github.com/ljpadam/CGPS

  • Publication . Preprint . Article . 2022
    Open Access
    Authors: 
    Yu, Lu; Pei, Shichao; Ding, Lizhong; Zhou, Jun; Li, Longfei; Zhang, Chuxu; Zhang, Xiangliang;
    Publisher: Association for the Advancement of Artificial Intelligence (AAAI)

    This paper studies learning node representations with graph neural networks (GNNs) for unsupervised scenario. Specifically, we derive a theoretical analysis and provide an empirical demonstration about the non-steady performance of GNNs over different graph datasets, when the supervision signals are not appropriately defined. The performance of GNNs depends on both the node feature smoothness and the locality of graph structure. To smooth the discrepancy of node proximity measured by graph topology and node feature, we proposed SAIL - a novel \underline{S}elf-\underline{A}ugmented graph contrast\underline{i}ve \underline{L}earning framework, with two complementary self-distilling regularization modules, \emph{i.e.}, intra- and inter-graph knowledge distillation. We demonstrate the competitive performance of SAIL on a variety of graph applications. Even with a single GNN layer, SAIL has consistently competitive or even better performance on various benchmark datasets, comparing with state-of-the-art baselines. Comment: Accepted by AAAI2022, 10 pages, 3 figures

  • Open Access English
    Authors: 
    Donald G. Dunagan; Shulin Zhang; Jixing Li; Shohini Bhattasali; Christophe Pallier; John Whitman; Yiming Yang; John Hale;
    Publisher: HAL CCSD
    Country: France

    AbstractOne aspect of natural language comprehension is understanding how many of what or whom a speaker is referring to. While previous work has documented the neural correlates of general number comprehension and quantity comparison, we investigate semantic number from a cross-linguistic perspective with the goal of identifying cortical regions involved in distinguishing plural from singular nouns. We use three fMRI datasets in which Chinese, French, and English native speakers listen to an audiobook of a children’s story in their native language. We select these three languages because they differ in their number semantics. While Chinese lacks nominal pluralization, French and English nouns are overtly marked for number. We find a number of known semantic processing regions in common, including dorsomedial prefrontal cortex and the pars orbitalis, in which cortical activation is greater for plural than singular nouns and posit a cross-linguistic role for number in semantic comprehension.

  • Publication . Article . Other literature type . Preprint . 2022 . Embargo End Date: 01 Jan 2019
    Open Access
    Authors: 
    Filipe Rodrigues; Nicola Ortelli; Michel Bierlaire; Francisco C. Pereira;
    Publisher: arXiv
    Countries: Switzerland, Denmark

    Specifying utility functions is a key step towards applying the discrete choice framework for understanding the behaviour processes that govern user choices. However, identifying the utility function specifications that best model and explain the observed choices can be a very challenging and time-consuming task. This paper seeks to help modellers by leveraging the Bayesian framework and the concept of automatic relevance determination (ARD), in order to automatically determine an optimal utility function specification from an exponentially large set of possible specifications in a purely data-driven manner. Based on recent advances in approximate Bayesian inference, a doubly stochastic variational inference is developed, which allows the proposed DCM-ARD model to scale to very large and high-dimensional datasets. Using semi-artificial choice data, the proposed approach is shown to very accurately recover the true utility function specifications that govern the observed choices. Moreover, when applied to real choice data, DCM-ARD is shown to be able discover high quality specifications that can outperform previous ones from the literature according to multiple criteria, thereby demonstrating its practical applicability. Comment: 21 pages, 2 figures, 11 tables

Advanced search in
Research products
arrow_drop_down
Searching FieldsTerms
Any field
arrow_drop_down
includes
arrow_drop_down
Include:
2,700 Research products, page 1 of 270
  • Open Access
    Authors: 
    Nadia Figueroa; Haiwei Dong; Abdulmotaleb El Saddik;
    Publisher: Association for Computing Machinery (ACM)
    Country: Switzerland

    We propose a 6D RGB-D odometry approach that finds the relative camera pose between consecutive RGB-D frames by keypoint extraction and feature matching both on the RGB and depth image planes. Furthermore, we feed the estimated pose to the highly accurate KinectFusion algorithm, which uses a fast ICP (Iterative Closest Point) to fine-tune the frame-to-frame relative pose and fuse the depth data into a global implicit surface. We evaluate our method on a publicly available RGB-D SLAM benchmark dataset by Sturm et al. The experimental results show that our proposed reconstruction method solely based on visual odometry and KinectFusion outperforms the state-of-the-art RGB-D SLAM system accuracy. Moreover, our algorithm outputs a ready-to-use polygon mesh (highly suitable for creating 3D virtual worlds) without any postprocessing steps.

  • Open Access
    Authors: 
    Christian Boudreault; Hichem Eleuch; Michael Hilke; Richard MacKenzie;
    Publisher: American Physical Society (APS)
    Project: NSERC

    One of the most challenging problems for the realization of a scalable quantum computer is to design a physical device that keeps the error rate for each quantum processing operation low. These errors can originate from the accuracy of quantum manipulation, such as the sweeping of a gate voltage in solid state qubits or the duration of a laser pulse in optical schemes. Errors also result from decoherence, which is often regarded as more crucial in the sense that it is inherent to the quantum system, being fundamentally a consequence of the coupling to the external environment. Grouping small collections of qubits into clusters with symmetries can protect parts of the calculation from decoherence. We use 4-level cores with a straightforward generalization of discrete rotational symmetry, omega-rotation invariance, to encode pairs of coupled qubits and universal 2-qubit logical gates. We include quantum errors as a main source of decoherence, and show that symmetry makes logical operations particularly resilient to untimely anisotropic qubit rotations. We propose a scalable scheme for universal quantum computation where cores play the role of quantum-computational transistors, quansistors. Initialization and readout are achieved by coupling to leads. The external leads are explicitly considered and are assumed to be the other main source of decoherence. We show that quansistors can be dynamically decoupled from the leads by tuning their internal parameters, giving them the versatility required to act as controllable quantum memory units. With this dynamical decoupling, logical operations within quansistors are also symmetry-protected from unbiased noise in their parameters. We identify technologies that could implement omega-rotation invariance. Many of our results can be generalized to higher-level omega-rotation-invariant systems, or adapted to clusters with other symmetries. Comment: 23 pages, 19 figures

  • Open Access
    Authors: 
    Montasir Qasymeh; Hichem Eleuch;
    Publisher: Springer Science and Business Media LLC

    Quantum microwave transmission is key to realizing modular superconducting quantum computers and distributed quantum networks. A large number of incoherent photons are thermally generated within the microwave frequency spectrum. The closeness of the transmitted quantum state to the source-generated quantum state at the input of the transmission link (measured by the transmission fidelity) degrades due to the presence of the incoherent photons. Hence, high-fidelity quantum microwave transmission has long been considered to be infeasible without refrigeration [3,4]. In this study, we propose a novel method for high-fidelity quantum microwave transmission using a room-temperature lossy waveguide. The proposed scheme consists of connecting two cryogenic nodes (i.e., a transmitter and a receiver) by the room-temperature lossy microwave waveguide. First, cryogenic preamplification is implemented prior to transmission. Second, at the receiver side, a cryogenic loop antenna is placed inside the output port of the waveguide and coupled to an LC harmonic oscillator located outside the waveguide. The loop antenna converts quantum microwave fields (which contain both signal and noise photons) to a quantum voltage across the coupled LC harmonic oscillator. The loop antenna detector at the receiver is designed to extensively suppress the induced photons across the LC oscillator. The signal transmittance is maintained intact by providing significant preamplification gain. Our calculations show that high-fidelity quantum transmission (i.e., more than 95%) is realized based on the proposed scheme for transmission distances reaching 100 m. Comment: 10pages; 6 figures

  • Open Access
    Authors: 
    Filippo Macchi; Eric Edsinger; Kirsten C. Sadler;
    Publisher: Cold Spring Harbor Laboratory

    Abstract Background Epigenetic regulatory mechanisms are divergent across the animal kingdom, yet these mechanisms are not well studied in non-model organisms. Unique features of cephalopods make them attractive for investigating behavioral, sensory, developmental, and regenerative processes, and recent studies have elucidated novel features of genome organization and gene and transposon regulation in these animals. However, it is not known how epigenetics regulates these interesting cephalopod features. We combined bioinformatic and molecular analysis of Octopus bimaculoides to investigate the presence and pattern of DNA methylation and examined the presence of DNA methylation and 3 histone post-translational modifications across tissues of three cephalopod species. Results We report a dynamic expression profile of the genes encoding conserved epigenetic regulators, including DNA methylation maintenance factors in octopus tissues. Levels of 5-methyl-cytosine in multiple tissues of octopus, squid, and bobtail squid were lower compared to vertebrates. Whole genome bisulfite sequencing of two regions of the brain and reduced representation bisulfite sequencing from a hatchling of O. bimaculoides revealed that less than 10% of CpGs are methylated in all samples, with a distinct pattern of 5-methyl-cytosine genome distribution characterized by enrichment in the bodies of a subset of 14,000 genes and absence from transposons. Hypermethylated genes have distinct functions and, strikingly, many showed similar expression levels across tissues while hypomethylated genes were silenced or expressed at low levels. Histone marks H3K27me3, H3K9me3, and H3K4me3 were detected at different levels across tissues of all species. Conclusions Our results show that the DNA methylation and histone modification epigenetic machinery is conserved in cephalopods, and that, in octopus, 5-methyl-cytosine does not decorate transposable elements, but is enriched on the gene bodies of highly expressed genes and could cooperate with the histone code to regulate tissue-specific gene expression.

  • Open Access English
    Authors: 
    Taushif Khan; Mahbubur Rahman; Ishfaq Ahmed; F. Al Ali; Puthen V. Jithesh; Nico Marr;
    Publisher: Frontiers Media SA

    Allelic diversity of human leukocyte antigen (HLA) class II genes may help maintain humoral immunity against infectious diseases. In this study, we investigated germline genetic variation in classical HLA class II genes and employed a systematic, unbiased approach to explore the relative contribution of this genetic variation in the antibody repertoire to various common pathogens. We leveraged a well-defined cohort of 800 adults representing the general Arab population in which genetic material is shared because of the high frequency of consanguineous unions. By applying a high-throughput method for large-scale antibody profiling to this well-defined cohort, we were able to dissect the overall effect of zygosity for classical HLA class II genes, as well as the effects associated with specific HLA class II alleles, haplotypes and genotypes, on the antimicrobial antibody repertoire breadth and antibody specificity with unprecedented resolution. Our population genetic studies revealed that zygosity of the classical HLA class II genes is a strong predictor of antibody responses to common human pathogens, suggesting that classical HLA class II gene heterozygosity confers a selective advantage. Moreover, we demonstrated that multiple HLA class II alleles can have additive effects on the antibody repertoire to common pathogens. We also identified associations of HLA-DRB1 genotypes with specific antigens. Our findings suggest that HLA class II gene polymorphisms confer specific humoral immunity against common pathogens, which may have contributed to the genetic diversity of HLA class II loci during hominine evolution.

  • Open Access
    Authors: 
    Tarun Dutta; Adrián Pérez-Salinas; Jasper Phua Sing Cheng; José Ignacio Latorre; Manas Mukherjee;
    Publisher: American Physical Society (APS)

    Quantum computers can provide solutions to classically intractable problems under specific and adequate conditions. However, current devices have only limited computational resources, and an effort is made to develop useful quantum algorithms under these circumstances. This work experimentally demonstrates that a single-qubit device can host a universal classifier. The quantum processor used in this work is based on ion traps, providing highly accurate control on small systems. The algorithm chosen is the re-uploading scheme, which can address general learning tasks. Ion traps suit the needs of accurate control required by re-uploading. In the experiment here presented, a set of non-trivial classification tasks are successfully carried. The training procedure is performed in two steps combining simulation and experiment. Final results are benchmarked against exact simulations of the same method and also classical algorithms, showing a competitive performance of the ion-trap quantum classifier. This work constitutes the first experimental implementation of a classification algorithm based on the re-uploading scheme. 13 pages, 11 figures, and 1 table

  • Publication . Preprint . Article . 2022
    Open Access
    Authors: 
    Yan, Yichao; Li, Jinpeng; Liao, Shengcai; Qin, Jie; Ni, Bingbing; Yang, Xiaokang; Shao, Ling;
    Publisher: Association for the Advancement of Artificial Intelligence (AAAI)

    Person search has recently emerged as a challenging task that jointly addresses pedestrian detection and person re-identification. Existing approaches follow a fully supervised setting where both bounding box and identity annotations are available. However, annotating identities is labor-intensive, limiting the practicability and scalability of current frameworks. This paper inventively considers weakly supervised person search with only bounding box annotations. We proposed to address this novel task by investigating three levels of context clues (i.e., detection, memory and scene) in unconstrained natural images. The first two are employed to promote local and global discriminative capabilities, while the latter enhances clustering accuracy. Despite its simple design, our CGPS achieves 80.0% in mAP on CUHK-SYSU, boosting the baseline model by 8.8%. Surprisingly, it even achieves comparable performance with several supervised person search models. Our code is available at https://github.com/ljpadam/CGPS

  • Publication . Preprint . Article . 2022
    Open Access
    Authors: 
    Yu, Lu; Pei, Shichao; Ding, Lizhong; Zhou, Jun; Li, Longfei; Zhang, Chuxu; Zhang, Xiangliang;
    Publisher: Association for the Advancement of Artificial Intelligence (AAAI)

    This paper studies learning node representations with graph neural networks (GNNs) for unsupervised scenario. Specifically, we derive a theoretical analysis and provide an empirical demonstration about the non-steady performance of GNNs over different graph datasets, when the supervision signals are not appropriately defined. The performance of GNNs depends on both the node feature smoothness and the locality of graph structure. To smooth the discrepancy of node proximity measured by graph topology and node feature, we proposed SAIL - a novel \underline{S}elf-\underline{A}ugmented graph contrast\underline{i}ve \underline{L}earning framework, with two complementary self-distilling regularization modules, \emph{i.e.}, intra- and inter-graph knowledge distillation. We demonstrate the competitive performance of SAIL on a variety of graph applications. Even with a single GNN layer, SAIL has consistently competitive or even better performance on various benchmark datasets, comparing with state-of-the-art baselines. Comment: Accepted by AAAI2022, 10 pages, 3 figures

  • Open Access English
    Authors: 
    Donald G. Dunagan; Shulin Zhang; Jixing Li; Shohini Bhattasali; Christophe Pallier; John Whitman; Yiming Yang; John Hale;
    Publisher: HAL CCSD
    Country: France

    AbstractOne aspect of natural language comprehension is understanding how many of what or whom a speaker is referring to. While previous work has documented the neural correlates of general number comprehension and quantity comparison, we investigate semantic number from a cross-linguistic perspective with the goal of identifying cortical regions involved in distinguishing plural from singular nouns. We use three fMRI datasets in which Chinese, French, and English native speakers listen to an audiobook of a children’s story in their native language. We select these three languages because they differ in their number semantics. While Chinese lacks nominal pluralization, French and English nouns are overtly marked for number. We find a number of known semantic processing regions in common, including dorsomedial prefrontal cortex and the pars orbitalis, in which cortical activation is greater for plural than singular nouns and posit a cross-linguistic role for number in semantic comprehension.

  • Publication . Article . Other literature type . Preprint . 2022 . Embargo End Date: 01 Jan 2019
    Open Access
    Authors: 
    Filipe Rodrigues; Nicola Ortelli; Michel Bierlaire; Francisco C. Pereira;
    Publisher: arXiv
    Countries: Switzerland, Denmark

    Specifying utility functions is a key step towards applying the discrete choice framework for understanding the behaviour processes that govern user choices. However, identifying the utility function specifications that best model and explain the observed choices can be a very challenging and time-consuming task. This paper seeks to help modellers by leveraging the Bayesian framework and the concept of automatic relevance determination (ARD), in order to automatically determine an optimal utility function specification from an exponentially large set of possible specifications in a purely data-driven manner. Based on recent advances in approximate Bayesian inference, a doubly stochastic variational inference is developed, which allows the proposed DCM-ARD model to scale to very large and high-dimensional datasets. Using semi-artificial choice data, the proposed approach is shown to very accurately recover the true utility function specifications that govern the observed choices. Moreover, when applied to real choice data, DCM-ARD is shown to be able discover high quality specifications that can outperform previous ones from the literature according to multiple criteria, thereby demonstrating its practical applicability. Comment: 21 pages, 2 figures, 11 tables

Send a message
How can we help?
We usually respond in a few hours.