Showing 1–15 of 15 results for author: Hager, J

Weight Re-Mapping for Variational Quantum Algorithms

Authors: Michael Kölle, Alessandro Giovagnoli, Jonas Stein, Maximilian Balthasar Mansky, Julian Hager, Tobias Rohe, Robert Müller, Claudia Linnhoff-Popien

Abstract: Inspired by the remarkable success of artificial neural networks across a broad spectrum of AI tasks, variational quantum circuits (VQCs) have recently seen an upsurge in quantum machine learning applications. The promising outcomes shown by VQCs, such as improved generalization and reduced parameter training requirements, are attributed to the robust algorithmic capabilities of quantum computing.… ▽ More Inspired by the remarkable success of artificial neural networks across a broad spectrum of AI tasks, variational quantum circuits (VQCs) have recently seen an upsurge in quantum machine learning applications. The promising outcomes shown by VQCs, such as improved generalization and reduced parameter training requirements, are attributed to the robust algorithmic capabilities of quantum computing. However, the current gradient-based training approaches for VQCs do not adequately accommodate the fact that trainable parameters (or weights) are typically used as angles in rotational gates. To address this, we extend the concept of weight re-mapping for VQCs, as introduced by Kölle et al. (2023). This approach unambiguously maps the weights to an interval of length $2πぱい$, mirroring data rescaling techniques in conventional machine learning that have proven to be highly beneficial in numerous scenarios. In our study, we employ seven distinct weight re-mapping functions to assess their impact on eight classification datasets, using variational classifiers as a representative example. Our results indicate that weight re-mapping can enhance the convergence speed of the VQC. We assess the efficacy of various re-mapping functions across all datasets and measure their influence on the VQC's average performance. Our findings indicate that weight re-mapping not only consistently accelerates the convergence of VQCs, regardless of the specific re-mapping function employed, but also significantly increases accuracy in certain cases. △ Less

Submitted 9 June, 2023; originally announced June 2023.

Introducing Reduced-Width QNNs, an AI-inspired Ansatz Design Pattern

Authors: Jonas Stein, Tobias Rohe, Francesco Nappi, Julian Hager, David Bucher, Maximilian Zorn, Michael Kölle, Claudia Linnhoff-Popien

Abstract: Variational Quantum Algorithms are one of the most promising candidates to yield the first industrially relevant quantum advantage. Being capable of arbitrary function approximation, they are often referred to as Quantum Neural Networks (QNNs) when being used in analog settings as classical Artificial Neural Networks (ANNs). Similar to the early stages of classical machine learning, known schemes… ▽ More Variational Quantum Algorithms are one of the most promising candidates to yield the first industrially relevant quantum advantage. Being capable of arbitrary function approximation, they are often referred to as Quantum Neural Networks (QNNs) when being used in analog settings as classical Artificial Neural Networks (ANNs). Similar to the early stages of classical machine learning, known schemes for efficient architectures of these networks are scarce. Exploring beyond existing design patterns, we propose a reduced-width circuit ansatz design, which is motivated by recent results gained in the analysis of dropout regularization in QNNs. More precisely, this exploits the insight, that the gates of overparameterized QNNs can be pruned substantially until their expressibility decreases. The results of our case study show, that the proposed design pattern can significantly reduce training time while maintaining the same result quality as the standard "full-width" design in the presence of noise. △ Less

Submitted 8 January, 2024; v1 submitted 8 June, 2023; originally announced June 2023.

Journal ref: Proceedings of the 16th International Conference on Agents and Artificial Intelligence - Volume 3. ICAART 2024

Improving Convergence for Quantum Variational Classifiers using Weight Re-Mapping

Authors: Michael Kölle, Alessandro Giovagnoli, Jonas Stein, Maximilian Balthasar Mansky, Julian Hager, Claudia Linnhoff-Popien

Abstract: In recent years, quantum machine learning has seen a substantial increase in the use of variational quantum circuits (VQCs). VQCs are inspired by artificial neural networks, which achieve extraordinary performance in a wide range of AI tasks as massively parameterized function approximators. VQCs have already demonstrated promising results, for example, in generalization and the requirement for fe… ▽ More In recent years, quantum machine learning has seen a substantial increase in the use of variational quantum circuits (VQCs). VQCs are inspired by artificial neural networks, which achieve extraordinary performance in a wide range of AI tasks as massively parameterized function approximators. VQCs have already demonstrated promising results, for example, in generalization and the requirement for fewer parameters to train, by utilizing the more robust algorithmic toolbox available in quantum computing. A VQCs' trainable parameters or weights are usually used as angles in rotational gates and current gradient-based training methods do not account for that. We introduce weight re-mapping for VQCs, to unambiguously map the weights to an interval of length $2πぱい$, drawing inspiration from traditional ML, where data rescaling, or normalization techniques have demonstrated tremendous benefits in many circumstances. We employ a set of five functions and evaluate them on the Iris and Wine datasets using variational classifiers as an example. Our experiments show that weight re-mapping can improve convergence in all tested settings. Additionally, we were able to demonstrate that weight re-mapping increased test accuracy for the Wine dataset by $10\%$ over using unmodified weights. △ Less

Submitted 15 February, 2023; v1 submitted 22 December, 2022; originally announced December 2022.

Comments: Accepted at ICAART 2023

Network analyses reveal negative link between changes in adipose tissue GDF15 and BMI during dietary-induced weight loss

Authors: Alyssa Imbert, Nathalie Vialaneix, Julien Marquis, Julie Vion, Aline Charpagne, Sylviane Metairon, Claire Laurens, Cedric Moro, Nathalie Boulet, Ondine Walter, Grégory Lefebvre, Jörg Hager, Dominique Langin, Wim Saris, Arne Astrup, Nathalie Viguerie, Armand Valsesia

Abstract: Adipose tissue (AT) transcriptome studies provide holistic pictures of adaptation to weight and related bioclinical settings changes. Objective To implement AT gene expression profiling and investigate the link between changes in bioclinical parameters and AT gene expression during 3 steps of a 2-phase dietary intervention (DI). Methods AT transcriptome profiling was obtained from sequencing 1051… ▽ More Adipose tissue (AT) transcriptome studies provide holistic pictures of adaptation to weight and related bioclinical settings changes. Objective To implement AT gene expression profiling and investigate the link between changes in bioclinical parameters and AT gene expression during 3 steps of a 2-phase dietary intervention (DI). Methods AT transcriptome profiling was obtained from sequencing 1051 samples, corresponding to 556 distinct individuals enrolled in a weight loss intervention (8-week low-calorie diet (LCD) at 800 kcal/day) followed with a 6-month ad libitum randomized DI. Transcriptome profiles obtained with QuantSeq sequencing were benchmarked against Illumina RNAseq. Reverse transcription quantitative polymerase chain reaction was used to further confirm associations. Cell specificity was assessed using freshly isolated cells and THP-1 cell line. Results During LCD, 5 modules were found, of which 3 included at least 1 bioclinical variable. Change in body mass index (BMI) connected with changes in mRNA level of genes with inflammatory response signature. In this module, change in BMI was negatively associated with changes in expression of genes encoding secreted protein (GDF15, CCL3, and SPP1). Through all phases of the DI, change in GDF15 was connected to changes in SPP1, CCL3, LIPA and CD68. Further characterization showed that these genes were specific to macrophages (with LIPA, CD68 and GDF15 expressed in anti-inflammatory macrophages) and GDF15 also expressed in preadipocytes. Conclusion Network analyses identified a novel AT feature with GDF15 upregulated with calorie restriction induced weight loss, concomitantly to macrophage markers. In AT, GDF15 was expressed in preadipocytes and macrophages where it was a hallmark of anti-inflammatory cells. △ Less

Submitted 17 March, 2022; originally announced March 2022.

Journal ref: Journal of Clinical Endocrinology and Metabolism, Endocrine Society, 2022, 107 (1), pp.e130-e142

GraspME -- Grasp Manifold Estimator

Authors: Janik Hager, Ruben Bauer, Marc Toussaint, Jim Mainprice

Abstract: In this paper, we introduce a Grasp Manifold Estimator (GraspME) to detect grasp affordances for objects directly in 2D camera images. To perform manipulation tasks autonomously it is crucial for robots to have such graspability models of the surrounding objects. Grasp manifolds have the advantage of providing continuously infinitely many grasps, which is not the case when using other grasp repres… ▽ More In this paper, we introduce a Grasp Manifold Estimator (GraspME) to detect grasp affordances for objects directly in 2D camera images. To perform manipulation tasks autonomously it is crucial for robots to have such graspability models of the surrounding objects. Grasp manifolds have the advantage of providing continuously infinitely many grasps, which is not the case when using other grasp representations such as predefined grasp points. For instance, this property can be leveraged in motion optimization to define goal sets as implicit surface constraints in the robot configuration space. In this work, we restrict ourselves to the case of estimating possible end-effector positions directly from 2D camera images. To this extend, we define grasp manifolds via a set of key points and locate them in images using a Mask R-CNN backbone. Using learned features allows generalizing to different view angles, with potentially noisy images, and objects that were not part of the training set. We rely on simulation data only and perform experiments on simple and complex objects, including unseen ones. Our framework achieves an inference speed of 11.5 fps on a GPU, an average precision for keypoint estimation of 94.5% and a mean pixel distance of only 1.29. This shows that we can estimate the objects very well via bounding boxes and segmentation masks as well as approximate the correct grasp manifold's keypoint coordinates. △ Less

Submitted 5 July, 2021; originally announced July 2021.

Comments: Accepted to RoMan 2021

Deep Workpiece Region Segmentation for Bin Picking

Authors: Muhammad Usman Khalid, Janik M. Hager, Werner Kraus, Marco F. Huber, Marc Toussaint

Abstract: For most industrial bin picking solutions, the pose of a workpiece is localized by matching a CADきゃど model to point cloud obtained from 3D sensor. Distinguishing flat workpieces from bottom of the bin in point cloud imposes challenges in the localization of workpieces that lead to wrong or phantom detections. In this paper, we propose a framework that solves this problem by automatically segmenting w… ▽ More For most industrial bin picking solutions, the pose of a workpiece is localized by matching a CADきゃど model to point cloud obtained from 3D sensor. Distinguishing flat workpieces from bottom of the bin in point cloud imposes challenges in the localization of workpieces that lead to wrong or phantom detections. In this paper, we propose a framework that solves this problem by automatically segmenting workpiece regions from non-workpiece regions in a point cloud data. It is done in real time by applying a fully convolutional neural network trained on both simulated and real data. The real data has been labelled by our novel technique which automatically generates ground truth labels for real point clouds. Along with real time workpiece segmentation, our framework also helps in improving the number of detected workpieces and estimating the correct object poses. Moreover, it decreases the computation time by approximately 1s due to a reduction of the search space for the object pose estimation. △ Less

Submitted 8 September, 2019; originally announced September 2019.

Comments: IEEE CASE 2019

A global-local approach for detecting hotspots in multiple-response regression

Authors: Hélène Ruffieux, Anthony C. Davison, Jörg Hager, Jamie Inshaw, Benjamin P. Fairfax, Sylvia Richardson, Leonardo Bottolo

Abstract: We tackle modelling and inference for variable selection in regression problems with many predictors and many responses. We focus on detecting hotspots, i.e., predictors associated with several responses. Such a task is critical in statistical genetics, as hotspot genetic variants shape the architecture of the genome by controlling the expression of many genes and may initiate decisive functional… ▽ More We tackle modelling and inference for variable selection in regression problems with many predictors and many responses. We focus on detecting hotspots, i.e., predictors associated with several responses. Such a task is critical in statistical genetics, as hotspot genetic variants shape the architecture of the genome by controlling the expression of many genes and may initiate decisive functional mechanisms underlying disease endpoints. Existing hierarchical regression approaches designed to model hotspots suffer from two limitations: their discrimination of hotspots is sensitive to the choice of top-level scale parameters for the propensity of predictors to be hotspots, and they do not scale to large predictor and response vectors, e.g., of dimensions $10^3-10^5$ in genetic applications. We address these shortcomings by introducing a flexible hierarchical regression framework that is tailored to the detection of hotspots and scalable to the above dimensions. Our proposal implements a fully Bayesian model for hotspots based on the horseshoe shrinkage prior. Its global-local formulation shrinks noise globally and hence accommodates the highly sparse nature of genetic analyses, while being robust to individual signals, thus leaving the effects of hotspots unshrunk. Inference is carried out using a fast variational algorithm coupled with a novel simulated annealing procedure that allows efficient exploration of multimodal distributions. △ Less

Submitted 15 May, 2020; v1 submitted 8 November, 2018; originally announced November 2018.

Efficient inference for genetic association studies with multiple outcomes

Authors: Hélène Ruffieux, Anthony C. Davison, Jörg Hager, Irina Irincheeva

Abstract: Combined inference for heterogeneous high-dimensional data is critical in modern biology, where clinical and various kinds of molecular data may be available from a single study. Classical genetic association studies regress a single clinical outcome on many genetic variants one by one, but there is an increasing demand for joint analysis of many molecular outcomes and genetic variants in order to… ▽ More Combined inference for heterogeneous high-dimensional data is critical in modern biology, where clinical and various kinds of molecular data may be available from a single study. Classical genetic association studies regress a single clinical outcome on many genetic variants one by one, but there is an increasing demand for joint analysis of many molecular outcomes and genetic variants in order to unravel functional interactions. Unfortunately, most existing approaches to joint modelling are either too simplistic to be powerful or are impracticable for computational reasons. Inspired by Richardson et al. (2010, Bayesian Statistics 9), we consider a sparse multivariate regression model that allows simultaneous selection of predictors and associated responses. As Markov chain Monte Carlo (MCMC) inference on such models can be prohibitively slow when the number of genetic variants exceeds a few thousand, we propose a variational inference approach which produces posterior information very close to that of MCMC inference, at a much reduced computational cost. Extensive numerical experiments show that our approach outperforms popular variable selection methods and tailored Bayesian procedures, dealing within hours with problems involving hundreds of thousands of genetic variants and tens to hundreds of clinical or molecular outcomes. △ Less

Submitted 20 March, 2017; v1 submitted 12 September, 2016; originally announced September 2016.

Journal ref: Biostatistics, 2017

Choice of the uncritical manifold for dilute polymer solutions

Authors: Andrea Ostendorf, Johannes S. Hager

Abstract: We discuss the dependence of the results of renormalized perturbation theory for dilute polymer solutions on the choice of the uncritical manifold where the perturbation series are evaluated. Special emphasis is given to the influence of polydispersity corrections on the results of one and two loop calculations. For monodisperse solutions we establish that after a Borel resummation the dependenc… ▽ More We discuss the dependence of the results of renormalized perturbation theory for dilute polymer solutions on the choice of the uncritical manifold where the perturbation series are evaluated. Special emphasis is given to the influence of polydispersity corrections on the results of one and two loop calculations. For monodisperse solutions we establish that after a Borel resummation the dependence on the choice of the uncritical manifold decreases when higher orders of the expansion are taken into account. △ Less

Submitted 21 December, 2004; originally announced December 2004.

Comments: 16pages,8figures, dedicated to Prof. Lothar Schaefer on the occasion of his 60th birthday

Scaling of demixing curves and crossover from critical to tricritical behavior in polymer solutions

Authors: J. S. Hager, M. A. Anisimov, J. V. Sengers, E. E. Gorodetskii

Abstract: In this paper we show that the virial expansion up to third order for the osmotic pressure of a dilute polymer solution, including first-order perturbative corrections to the virial coefficients, allows for a scaling description of phase-separation data for polymer solutions in reduced variables. This scaling description provides a method to estimate the Theta-temperature, where demixing occurs… ▽ More In this paper we show that the virial expansion up to third order for the osmotic pressure of a dilute polymer solution, including first-order perturbative corrections to the virial coefficients, allows for a scaling description of phase-separation data for polymer solutions in reduced variables. This scaling description provides a method to estimate the Theta-temperature, where demixing occurs in the limit of vanishing polymer volume fraction $φふぁい$ and infinite chain-length $N$, without explicit assumptions concerning the chain-length dependence of the critical parameters $T_c$ and $φふぁい_c$. The scaling incorporates three limiting regimes: the Ising limit asymptotically close to the critical point of phase separation, the pure solvent limit and, the tricritical limit for the polymer-rich phase asymptotically close to the Theta point. We incorporate the effects of critical and tricritical fluctuations on the coexistence curve scaling by using renormalization-group methods. We present a detailed comparison with experimental and simulation data for coexistence curves and compare our estimates for the Theta-temperatures of several systems with those obtained from different extrapolation schemes. △ Less

Submitted 24 June, 2002; originally announced June 2002.

Comments: revtex,18 pages,10 eps figures, submitted to J. Chem. Phys

2D Crystal Shapes, Droplet Condensation and Exponential Slowing Down in Simulations of First-Order Phase Transitions

Authors: Thomas Neuhaus, Johannes S. Hager

Abstract: Multicanonical ensemble simulations for the simulation of first-order phase transitions suffer from exponential slowing down. Monte Carlo autocorrelation times diverge exponentially with free energy barriers $ΔでるたF$, which in $L^d$ boxes grow as $L^{d-1}$. We exemplify the situation in a study of the 2D Ising-model at temperature $T/T_c=0.63$ for two different lattice manifolds, toroidal lattices a… ▽ More Multicanonical ensemble simulations for the simulation of first-order phase transitions suffer from exponential slowing down. Monte Carlo autocorrelation times diverge exponentially with free energy barriers $ΔでるたF$, which in $L^d$ boxes grow as $L^{d-1}$. We exemplify the situation in a study of the 2D Ising-model at temperature $T/T_c=0.63$ for two different lattice manifolds, toroidal lattices and surfaces of cubes. For both geometries the effect is caused by discontinuous droplet shape transitions between various classical crystal shapes obeying geometrical constraints. We use classical droplet theory and numerical simulations to calculate transition points and barrier heights. On toroidal lattices we determine finite size corrections to the droplet free energy, which are given by a linear combination of Gibbs-Thomson corrections, capillary wave fluctuation corrections, constant terms and logarithmic terms in the droplet volume. Tolman corrections are absent. In addition, we study the finite size effects on the condensation phase transition, which occurs in infinite systems at the Onsager value of the magnetization. We find that this transition is of discontinuous order also. A combination of classical droplet theory and Gibbs-Thomson corrections yields a fair description for the transition point and for the droplet size discontinuity for large droplets. We also estimate the nucleation barrier that has to be surmounted in the formation of the stable droplet at coexistence. △ Less

Submitted 14 April, 2003; v1 submitted 17 January, 2002; originally announced January 2002.

Comments: 39 pages, 21 figures, version accepted by J. Stat. Phys

Journal ref: J.Stat.Phys. 113 (1-2) 47-83 Oct 2003

Phase Coherence and Control of Stored Photonic Information

Authors: A. Mair, J. Hager, D. F. Phillips, R. L. Walsworth, M. D. Lukin

Abstract: We report the demonstration of phase coherence and control for the recently developed "light storage" technique. Specifically, we use a pulsed magnetic field to vary the phase of atomic spin excitations which result from the deceleration and storing of a light pulse in warm Rb vapor. We then convert the spin excitations back into light and detect the resultant phase shift in an optical interfero… ▽ More We report the demonstration of phase coherence and control for the recently developed "light storage" technique. Specifically, we use a pulsed magnetic field to vary the phase of atomic spin excitations which result from the deceleration and storing of a light pulse in warm Rb vapor. We then convert the spin excitations back into light and detect the resultant phase shift in an optical interferometric measurement. The coherent storage of photon states in matter is essential for the practical realization of many basic concepts in quantum information processing. △ Less

Submitted 9 August, 2001; originally announced August 2001.

Comments: 5 pages, 3 figures. Submitted to Phys. Rev. Lett

Fluctuations of steps on crystal surfaces

Authors: W. Selke, F. Szalma, J. S. Hager

Abstract: Fluctuations of isolated and pairs of ascending steps of monoatomic height are studied in the framework of SOS models, using mainly Monte Carlo techniques. Below the roughening transistion of the surface, the profiles of long steps show the same scaling features for terrace and surface diffusion. For a pair of short steps, their separation distance is found to grow as $t^{1/3}$ at late stages. A… ▽ More Fluctuations of isolated and pairs of ascending steps of monoatomic height are studied in the framework of SOS models, using mainly Monte Carlo techniques. Below the roughening transistion of the surface, the profiles of long steps show the same scaling features for terrace and surface diffusion. For a pair of short steps, their separation distance is found to grow as $t^{1/3}$ at late stages. Above roughening, simulational data on surface diffusion agree well with the classical continuum theory of Mullins. △ Less

Submitted 24 August, 2001; originally announced August 2001.

Comments: 4 pages, 2 eps figures

Theta-point behavior of diluted polymer solutions: Can one observe the universal logarithmic corrections predicted by field theory?

Authors: Johannes Hager, Lothar Sch"afer

Abstract: In recent large scale Monte-Carlo simulations of various models of Theta-point polymers in three dimensions Grassberger and Hegger found logarithmic corrections to mean field theory with amplitudes much larger than the universal amplitudes of the leading logarithmic corrections calculated by Duplantier in the framework of tricritical O(n) field theory. To resolve this issue we calculate the univ… ▽ More In recent large scale Monte-Carlo simulations of various models of Theta-point polymers in three dimensions Grassberger and Hegger found logarithmic corrections to mean field theory with amplitudes much larger than the universal amplitudes of the leading logarithmic corrections calculated by Duplantier in the framework of tricritical O(n) field theory. To resolve this issue we calculate the universal subleading correction of field theory, which turns out to be of the same order of magnitude as the leading correction for all chain lengths available in present days simulations. Borel resummation of the renormalization group flow equations also shows the presence of such large corrections. This suggests that the published simulations did not reach the asymptotic regime. To further support this view, we present results of Monte-Carlo simulations on a Domb-Joyce like model of weakly interacting random walks. Again the results cannot be explained by keeping only the leading corrections, but are in fair accord with our full theoretical result. The corrections found for the Domb-Joyce model are much smaller than those for other models, which clearly shows that the effective corrections are not yet in the asymptotic regime. All together our findings show that the existing simulations of Theta-polymers are compatible with tricritical field theory since the crossover to the asymptotic regime is very slow. Similar results were found earlier for self avoiding walks at their upper critical dimension d=4. △ Less

Submitted 31 August, 1999; originally announced August 1999.

Comments: 15 pages,6 figures

Journal ref: PRE 60, 2071 (1999)

Relaxation of Surface Profiles by Evaporation Dynamics

Authors: Johannes Hager

Abstract: We present simulations of the relaxation towards equilibrium of one dimensional steps and sinusoidal grooves imprinted on a surface below its roughening transition. We use a generalization of the hypercube stacking model of Forrest and Tang, that allows for temperature dependent next-nearest-neighbor interactions. For the step geometry the results at T=0 agree well with the t^(1/4) prediction of… ▽ More We present simulations of the relaxation towards equilibrium of one dimensional steps and sinusoidal grooves imprinted on a surface below its roughening transition. We use a generalization of the hypercube stacking model of Forrest and Tang, that allows for temperature dependent next-nearest-neighbor interactions. For the step geometry the results at T=0 agree well with the t^(1/4) prediction of continuum theory for the spreading of the step. In the case of periodic profiles we modify the mobility for the tips of the profile and find the approximate solution of the resulting free boundary problem to be in reasonable agreement with the T=0 simulations. △ Less

Submitted 18 July, 1997; originally announced July 1997.

Comments: 6 pages, Revtex, 5 Postscript figures, to appear in PRB 15, October 1997