Showing 1–50 of 255 results for author: Koch, R

Constructive Holography

Authors: Robert de Mello Koch, Hendrik J. R. Van Zyl

Abstract: We consider the collective field theory description of the singlet sector of a free and massless matrix field in $d$ dimensions. The $k$-local collective fields are functions of $(d-1)k+1$ coordinates. We provide a map between the collective fields and fields in the dual gravitational theory defined on AdS$_{d+1}$ spacetime. The coordinates of the collective field have a natural interpretation: th… ▽ More We consider the collective field theory description of the singlet sector of a free and massless matrix field in $d$ dimensions. The $k$-local collective fields are functions of $(d-1)k+1$ coordinates. We provide a map between the collective fields and fields in the dual gravitational theory defined on AdS$_{d+1}$ spacetime. The coordinates of the collective field have a natural interpretation: the $k$-local collective field is a field defined on an AdS$_{d+1}\times$S$^{k-1}\times$S$^{(d-2)(k-2)}\times$S$^{d-3}$ spacetime. The modes of a harmonic expansion on the S$^{k-1}\times$S$^{(d-2)(k-2)}\times$S$^{d-3}$ portion of the spacetime leads to the spinning bulk fields of the dual gravity theory. △ Less

Submitted 26 June, 2024; originally announced June 2024.

Cross-Platform Autonomous Control of Minimal Kitaev Chains

Authors: David van Driel, Rouven Koch, Vincent P. M. Sietses, Sebastiaan L. D. ten Haaf, Chun-Xiao Liu, Francesco Zatelli, Bart Roovers, Alberto Bordin, Nick van Loo, Guanzhong Wang, Jan Cornelis Wolff, Grzegorz P. Mazur, Tom Dvir, Ivan Kulesh, Qingzhen Wang, A. Mert Bozkurt, Sasa Gazibegovic, Ghada Badawy, Erik P. A. M. Bakkers, Michael Wimmer, Srijit Goswami, Jose L. Lado, Leo P. Kouwenhoven, Eliska Greplova

Abstract: Contemporary quantum devices are reaching new limits in size and complexity, allowing for the experimental exploration of emergent quantum modes. However, this increased complexity introduces significant challenges in device tuning and control. Here, we demonstrate autonomous tuning of emergent Majorana zero modes in a minimal realization of a Kitaev chain. We achieve this task using cross-platfor… ▽ More Contemporary quantum devices are reaching new limits in size and complexity, allowing for the experimental exploration of emergent quantum modes. However, this increased complexity introduces significant challenges in device tuning and control. Here, we demonstrate autonomous tuning of emergent Majorana zero modes in a minimal realization of a Kitaev chain. We achieve this task using cross-platform transfer learning. First, we train a tuning model on a theory model. Next, we retrain it using a Kitaev chain realization in a two-dimensional electron gas. Finally, we apply this model to tune a Kitaev chain realized in quantum dots coupled through a semiconductor-superconductor section in a one-dimensional nanowire. Utilizing a convolutional neural network, we predict the tunneling and Cooper pair splitting rates from differential conductance measurements, employing these predictions to adjust the electrochemical potential to a Majorana sweet spot. The algorithm successfully converges to the immediate vicinity of a sweet spot (within 1.5 mV in 67.6% of attempts and within 4.5 mV in 80.9% of cases), typically finding a sweet spot in 45 minutes or less. This advancement is a step** stone towards autonomous tuning of emergent modes in interacting systems, and towards foundational tuning machine learning models that can be deployed across a range of experimental platforms. △ Less

Submitted 7 May, 2024; originally announced May 2024.

Reconstructing the spacetime dual to a free matrix

Authors: Robert de Mello Koch, Pratik Roy, Hendrik J. R. Van Zyl

Abstract: In this paper we consider the collective field theory description of the singlet sector of a free matrix field in 2+1 dimensions. This necessarily involves the study of $k$-local collective fields, which are functions of $2k+1$ coordinates. We argue that these coordinates have a natural interpretation: the $k$-local collective field is a field defined on an AdS$_4\times$S$^{k-2}\times$S$^{k-1}$ sp… ▽ More In this paper we consider the collective field theory description of the singlet sector of a free matrix field in 2+1 dimensions. This necessarily involves the study of $k$-local collective fields, which are functions of $2k+1$ coordinates. We argue that these coordinates have a natural interpretation: the $k$-local collective field is a field defined on an AdS$_4\times$S$^{k-2}\times$S$^{k-1}$ spacetime. The modes of a harmonic expansion on the S$^{k-2}\times$S$^{k-1}$ portion of the spacetime leads to the spinning bulk fields of the dual gravity theory. △ Less

Submitted 7 May, 2024; originally announced May 2024.

Holography of a single free matrix

Authors: Robert de Mello Koch, Pratik Roy, Hendrik J. R. Van Zyl

Abstract: In this paper we consider the collective field theory description of a single free massless scalar matrix theory in 2+1 dimensions. The collective fields are given by $k$-local operators obtained by tracing a product of $k$-matrices. For $k=2$ and $k=3$ we argue that the collective field packages the fields associated to a single and two Regge trajectories respectively. We also determine the coord… ▽ More In this paper we consider the collective field theory description of a single free massless scalar matrix theory in 2+1 dimensions. The collective fields are given by $k$-local operators obtained by tracing a product of $k$-matrices. For $k=2$ and $k=3$ we argue that the collective field packages the fields associated to a single and two Regge trajectories respectively. We also determine the coordinate transformation between the coordinates of the collective field theory and the bulk AdS space time. This is used to verify that the bulk equations of motion holds in the collective field theory description. △ Less

Submitted 28 March, 2024; originally announced March 2024.

Comments: 1+31 pages

Bias Control and Linearization of the Transfer Function of Electro-optic and Acousto-optic Modulators

Authors: Clemens Neumüller, Frank Obernosterer, Raimund Meyer, Robert Koch, Gerd Kilian

Abstract: In several types of quantum computers light is one of the main tools to control both the position and the quantum state of the atoms used for computing. In practical systems laser light is applied to manipulate quantum states of qubits in the desired way. Beside physical effects like decoherence and quantum noise the precision of qubit manipulation has a significant impact on the achievable quantu… ▽ More In several types of quantum computers light is one of the main tools to control both the position and the quantum state of the atoms used for computing. In practical systems laser light is applied to manipulate quantum states of qubits in the desired way. Beside physical effects like decoherence and quantum noise the precision of qubit manipulation has a significant impact on the achievable quantum computing error rate. One of the key optical components beside the laser is the optical modulator, which modulates or switches a constant power laser light in order to provide light pulses or pulse sequences with a desired envelope. Acousto-optic (AOM) and electro-optic (EOM) modulators can be applied, which are both voltage controlled. However, there is neither a simple linear relationship between their control signal and the precise modulator output, nor can they be considered to have time-invariant characteristics. The aim of this paper is to describe techniques to generate AOM and EOM control signals in such a way that almost arbitrary target output waveforms (i. e. optical power versus time) are achieved with high accuracy. △ Less

Submitted 15 March, 2024; originally announced March 2024.

Bilocal holography and locality in the bulk

Authors: Robert de Mello Koch, Garreth Kemp, Hendrik J. R. Van Zyl

Abstract: Bilocal holography provides a constructive approach to the vector model/higher spin gravity duality. It has two ingredients: a change of field variables and a change of space time coordinates. The change of field variables ensures that the loop expansion parameter becomes ${1\over N}$. The change of coordinates solves the Clebsch-Gordan problem of moving from the tensor product basis (in which the… ▽ More Bilocal holography provides a constructive approach to the vector model/higher spin gravity duality. It has two ingredients: a change of field variables and a change of space time coordinates. The change of field variables ensures that the loop expansion parameter becomes ${1\over N}$. The change of coordinates solves the Clebsch-Gordan problem of moving from the tensor product basis (in which the collective bilocal field is written) to the direct sum basis (appropriate for the description of the gravity fields). We argue that the change of space time coordinates can be deduced by requiring that operators constructed in the bilocal collective field theory are dual to local operators in the AdS bulk. △ Less

Submitted 11 June, 2024; v1 submitted 12 March, 2024; originally announced March 2024.

Comments: 1+16 pages

Topological rejection of noise by non-local quantum skyrmions

Authors: Pedro Ornelas, Isaac Nape, Robert De Mello Koch, Andrew Forbes

Abstract: An open challenge in the context of quantum information processing and communication is improving the robustness of quantum information to environmental contributions of noise, a severe hindrance in real-world scenarios. Here we create quantum skyrmionic topologies across two degrees of freedom in two entangled photons and show that the topological invariant remains resilient even as the purity of… ▽ More An open challenge in the context of quantum information processing and communication is improving the robustness of quantum information to environmental contributions of noise, a severe hindrance in real-world scenarios. Here we create quantum skyrmionic topologies across two degrees of freedom in two entangled photons and show that the topological invariant remains resilient even as the purity of the state decreases due to environmental noise, collapsing only once the state is maximally mixed. We discuss how such noise perturbations can be considered a smooth deformation of the state, and provide a full theoretical treatment to support the experimental findings. Our approach can be viewed as a digitization of quantum information, holding exciting promise for intrinsic quantum resilience through topology. △ Less

Submitted 4 March, 2024; originally announced March 2024.

Mind the Exit Pupil Gap: Revisiting the Intrinsics of a Standard Plenoptic Camera

Authors: Tim Michels, Daniel Mäckelmann, Reinhard Koch

Abstract: Among the common applications of plenoptic cameras are depth reconstruction and post-shot refocusing. These require a calibration relating the camera-side light field to that of the scene. Numerous methods with this goal have been developed based on thin lens models for the plenoptic camera's main lens and microlenses. Our work addresses the often-overlooked role of the main lens exit pupil in the… ▽ More Among the common applications of plenoptic cameras are depth reconstruction and post-shot refocusing. These require a calibration relating the camera-side light field to that of the scene. Numerous methods with this goal have been developed based on thin lens models for the plenoptic camera's main lens and microlenses. Our work addresses the often-overlooked role of the main lens exit pupil in these models and specifically in the decoding process of standard plenoptic camera (SPC) images. We formally deduce the connection between the refocusing distance and the resampling parameter for the decoded light field and provide an analysis of the errors that arise when the exit pupil is not considered. In addition, previous work is revisited with respect to the exit pupil's role and all theoretical results are validated through a ray-tracing-based simulation. With the public release of the evaluated SPC designs alongside our simulation and experimental data we aim to contribute to a more accurate and nuanced understanding of plenoptic camera optics. △ Less

Submitted 5 April, 2024; v1 submitted 20 February, 2024; originally announced February 2024.

Comments: 29 pages, 16 figures, Accepted for publication in MDPI Sensors, Special Issue 'Short-Range Optical 3D Scanning and 3D Data Processing '

A time-delayed model for active mode-locking

Authors: Elias R. Koch, Svetlana V. Gurevich, Julien Javaloyes

Abstract: We propose a first-principle time-delayed model for the study of active mode-locking that is devoid of restriction regarding the values of the round-trip gain and losses. It allows us to access the typical regimes encountered in semiconductor lasers and to perform an extended bifurcation analysis. Close to the harmonic resonances and to the lasing threshold, we recover the Hermite-Gauss solutions.… ▽ More We propose a first-principle time-delayed model for the study of active mode-locking that is devoid of restriction regarding the values of the round-trip gain and losses. It allows us to access the typical regimes encountered in semiconductor lasers and to perform an extended bifurcation analysis. Close to the harmonic resonances and to the lasing threshold, we recover the Hermite-Gauss solutions. However, the presence of the linewidth enhancement factor induces complex regimes in which even the fundamental solution becomes unstable. We discover a global bifurcation scenario in which a single pulse can jump, over a slow time scale, between the different minima of the modulation potential. Finally, we derive a Haus master equation close to the lasing threshold which shows a good agreement with the original time-delayed model. △ Less

Submitted 10 January, 2024; originally announced January 2024.

HoLLiE C -- A Multifunctional Bimanual Mobile Robot Supporting Versatile Care Applications

Authors: Lea Steffen, Martin Schulze, Christian Eichmann, Robin Koch, Andreas Hermann, Rosa Frietsch Mussulin, Friedrich Graaf, Robert Wilbrandt, Marvin Große Besselmann, Arne Roennau, Rüdiger Dillmann

Abstract: Care robotics as a research field has developed a lot in recent years, driven by the rapidly increasing need for it. However, these technologies are mostly limited to a very concrete and usually relatively simple use case. The bimanual robot House of Living Labs intelligent Escort (HoLLiE) includes an omnidirectional mobile platform. This paper presents how HoLLiE is adapted, by flexible software… ▽ More Care robotics as a research field has developed a lot in recent years, driven by the rapidly increasing need for it. However, these technologies are mostly limited to a very concrete and usually relatively simple use case. The bimanual robot House of Living Labs intelligent Escort (HoLLiE) includes an omnidirectional mobile platform. This paper presents how HoLLiE is adapted, by flexible software and hardware modules, for different care applications. The design goal of HoLLiE was to be human-like but abstract enough to ensure a high level of acceptance, which is very advantageous for its use in hospitals. After a short retrospect of previous generations of HoLLiE, it is highlighted how the current version is equipped with a variety of additional sensors and actuators to allow a wide range of possible applications. Then, the software stack of HoLLiE is depicted, with the focus on navigation and force sensitive intention recognition. △ Less

Submitted 11 December, 2023; originally announced December 2023.

Comments: 18th international conference on Intelligent Autonomous Systems (IAS18 - 2023)

Transfer of Reinforcement Learning-Based Controllers from Model- to Hardware-in-the-Loop

Authors: Mario Picerno, Lucas Koch, Kevin Badalian, Marius Wegener, Joschka Schaub, Charles Robert Koch, Jakob Andert

Abstract: The process of develo** control functions for embedded systems is resource-, time-, and data-intensive, often resulting in sub-optimal cost and solutions approaches. Reinforcement Learning (RL) has great potential for autonomously training agents to perform complex control tasks with minimal human intervention. Due to costly data generation and safety constraints, however, its application is mos… ▽ More The process of develo** control functions for embedded systems is resource-, time-, and data-intensive, often resulting in sub-optimal cost and solutions approaches. Reinforcement Learning (RL) has great potential for autonomously training agents to perform complex control tasks with minimal human intervention. Due to costly data generation and safety constraints, however, its application is mostly limited to purely simulated domains. To use RL effectively in embedded system function development, the generated agents must be able to handle real-world applications. In this context, this work focuses on accelerating the training process of RL agents by combining Transfer Learning (TL) and X-in-the-Loop (XiL) simulation. For the use case of transient exhaust gas re-circulation control for an internal combustion engine, use of a computationally cheap Model-in-the-Loop (MiL) simulation is made to select a suitable algorithm, fine-tune hyperparameters, and finally train candidate agents for the transfer. These pre-trained RL agents are then fine-tuned in a Hardware-in-the-Loop (HiL) system via TL. The transfer revealed the need for adjusting the reward parameters when advancing to real hardware. Further, the comparison between a purely HiL-trained and a transferred agent showed a reduction of training time by a factor of 5.9. The results emphasize the necessity to train RL agents with real hardware, and demonstrate that the maturity of the transferred policies affects both training time and performance, highlighting the strong synergies between TL and XiL simulation. △ Less

Submitted 25 October, 2023; originally announced October 2023.

Introducing a Deep Neural Network-based Model Predictive Control Framework for Rapid Controller Implementation

Authors: David C. Gordon, Alexander Winkler, Julian Bedei, Patrick Schaber, Jakob Andert, Charles R. Koch

Abstract: Model Predictive Control (MPC) provides an optimal control solution based on a cost function while allowing for the implementation of process constraints. As a model-based optimal control technique, the performance of MPC strongly depends on the model used where a trade-off between model computation time and prediction performance exists. One solution is the integration of MPC with a machine learn… ▽ More Model Predictive Control (MPC) provides an optimal control solution based on a cost function while allowing for the implementation of process constraints. As a model-based optimal control technique, the performance of MPC strongly depends on the model used where a trade-off between model computation time and prediction performance exists. One solution is the integration of MPC with a machine learning (ML) based process model which are quick to evaluate online. This work presents the experimental implementation of a deep neural network (DNN) based nonlinear MPC for Homogeneous Charge Compression Ignition (HCCI) combustion control. The DNN model consists of a Long Short-Term Memory (LSTM) network surrounded by fully connected layers which was trained using experimental engine data and showed acceptable prediction performance with under 5% error for all outputs. Using this model, the MPC is designed to track the Indicated Mean Effective Pressure (IMEP) and combustion phasing trajectories, while minimizing several parameters. Using the acados software package to enable the real-time implementation of the MPC on an ARM Cortex A72, the optimization calculations are completed within 1.4 ms. The external A72 processor is integrated with the prototy** engine controller using a UDP connection allowing for rapid experimental deployment of the NMPC. The IMEP trajectory following of the developed controller was excellent, with a root-mean-square error of 0.133 bar, in addition to observing process constraints. △ Less

Submitted 12 October, 2023; originally announced October 2023.

Comments: Submitted to 2024 American Control Conference (ACC), July 8-12, 2024 in Toronto, Canada. ACC is the annual conference of the American Automatic Control Council (AACC), the U.S. national member organization of the International Federation for Automatic Control (IFAC)

Gravitational dynamics from collective field theory

Authors: Robert de Mello Koch

Abstract: We consider the relevance of a collective field theory description for the AdS/CFT correspondence. Collective field theory performs a systematic reorganization of the degrees of freedom of a (non-gravitational) field theory, replacing the original loop expansion parameter $\hbar$ with $1/N$. Collective fields are over complete signalling a redundancy inherent in the theory. We propose that this ov… ▽ More We consider the relevance of a collective field theory description for the AdS/CFT correspondence. Collective field theory performs a systematic reorganization of the degrees of freedom of a (non-gravitational) field theory, replacing the original loop expansion parameter $\hbar$ with $1/N$. Collective fields are over complete signalling a redundancy inherent in the theory. We propose that this over completeness is the mechanism by which one arrives at a holographic description, to be identified with the gravity dual. We find evidence for this by studying the redundancy of the collective field theory, showing that degrees of freedom in the bulk can be expressed as a linear combination of degrees of freedom contained in an arbitrarily small neighbourhood of the boundary. △ Less

Submitted 20 September, 2023; originally announced September 2023.

Comments: 24+1 pages

Discovery of interlayer plasmon polaron in graphene/WS$_2$ heterostructures

Authors: Søren Ulstrup, Yann in 't Veld, Jill A. Miwa, Alfred J. H. Jones, Kathleen M. McCreary, Jeremy T. Robinson, Berend T. Jonker, Simranjeet Singh, Roland J. Koch, Eli Rotenberg, Aaron Bostwick, Chris Jozwiak, Malte Rösner, Jyoti Katoch

Abstract: Harnessing electronic excitations involving coherent coupling to bosonic modes is essential for the design and control of emergent phenomena in quantum materials [1]. In situations where charge carriers induce a lattice distortion due to the electron-phonon interaction, the conducting states get "dressed". This leads to the formation of polaronic quasiparticles that dramatically impact charge tran… ▽ More Harnessing electronic excitations involving coherent coupling to bosonic modes is essential for the design and control of emergent phenomena in quantum materials [1]. In situations where charge carriers induce a lattice distortion due to the electron-phonon interaction, the conducting states get "dressed". This leads to the formation of polaronic quasiparticles that dramatically impact charge transport, surface reactivity, thermoelectric and optical properties, as observed in a variety of crystals and interfaces composed of polar materials [2-6]. Similarly, when oscillations of the charge density couple to conduction electrons the more elusive plasmon polaron emerges [7], which has been detected in electron-doped semiconductors [8-10]. However, the exploration of polaronic effects on low energy excitations is still in its infancy in two-dimensional (2D) materials. Here, we present the discovery of an interlayer plasmon polaron in heterostructures composed of graphene on top of SL WS$_2$. By using micro-focused angle-resolved photoemission spectroscopy (microARPES) during in situ do** of the top graphene layer, we observe a strong quasiparticle peak accompanied by several carrier density-dependent shake-off replicas around the SL WS$_2$ conduction band minimum (CBM). Our results are explained by an effective many-body model in terms of a coupling between SL WS$_2$ conduction electrons and graphene plasmon modes. It is important to take into account the presence of such interlayer collective modes, as they have profound consequences for the electronic and optical properties of heterostructures that are routinely explored in many device architectures involving 2D transition metal dichalcogenides (TMDs) [11-15]. △ Less

Submitted 31 August, 2023; originally announced August 2023.

Comments: 25 pages, 9 figures including Supporting Information

Square Waves and Bykov T-points in a Delay Algebraic Model for the Kerr-Gires-Tournois Interferometer

Authors: Mina Stöhr, Elias R. Koch, Julien Javaloyes, Svetlana V. Gurevich, Matthias Wolfrum

Abstract: We study theoretically the mechanisms of square wave formation of a vertically emitting micro-cavity operated in the Gires-Tournois regime that contains a Kerr medium and that is subjected to strong time-delayed optical feedback and detuned optical injection. We show that in the limit of large delay, square wave solutions of the time-delayed system can be treated as relative homoclinic solutions o… ▽ More We study theoretically the mechanisms of square wave formation of a vertically emitting micro-cavity operated in the Gires-Tournois regime that contains a Kerr medium and that is subjected to strong time-delayed optical feedback and detuned optical injection. We show that in the limit of large delay, square wave solutions of the time-delayed system can be treated as relative homoclinic solutions of an equation with an advanced argument. Based on this, we use concepts of classical homoclinic bifurcation theory to study different types of square wave solutions. In particular, we unveil the mechanisms behind the collapsed snaking scenario of square waves and explain the formation of complex-shaped multistable square wave solutions through a Bykov T-point. Finally we relate the position of the T-point to the position of the Maxwell point in the original time-delayed system. △ Less

Submitted 23 August, 2023; originally announced August 2023.

Hamiltonian learning with real-space impurity tomography in topological moire superconductors

Authors: Maryam Khosravian, Rouven Koch, Jose L. Lado

Abstract: Extracting Hamiltonian parameters from available experimental data is a challenge in quantum materials. In particular, real-space spectroscopy methods such as scanning tunneling spectroscopy allow probing electronic states with atomic resolution, yet even in those instances extracting effective Hamiltonian is an open challenge. Here we show that impurity states in modulated systems provide a promi… ▽ More Extracting Hamiltonian parameters from available experimental data is a challenge in quantum materials. In particular, real-space spectroscopy methods such as scanning tunneling spectroscopy allow probing electronic states with atomic resolution, yet even in those instances extracting effective Hamiltonian is an open challenge. Here we show that impurity states in modulated systems provide a promising approach to extracting non-trivial Hamiltonian parameters of a quantum material. We show that by combining the real-space spectroscopy of different impurity locations in a moire topological superconductor, modulations of exchange and superconducting parameters can be inferred via machine learning. We demonstrate our strategy with a physically-inspired harmonic expansion combined with a fully-connected neural network that we benchmark against a conventional convolutional architecture. We show that while both approaches allow extracting exchange modulations, only the former approach allows inferring the features of the superconducting order. Our results demonstrate the potential of machine learning methods to extract Hamiltonian parameters by real-space impurity spectroscopy as local probes of a topological state. △ Less

Submitted 21 January, 2024; v1 submitted 22 August, 2023; originally announced August 2023.

Comments: 11 pages, 8 figures

Journal ref: Journal of Physics: Materials, 7 015012 (2024)

Nature of the current-induced insulator-to-metal transition in Ca$_2$RuO$_4$ as revealed by transport-ARPES

Authors: Cissy T Suen, Igor Marković, Marta Zonno, Sergey Zhdanovich, Na-Hyun Jo, Michael Schmid, Philipp Hansmann, Pascal Puphal, Katrin Fürsich, Valentin Zimmerman, Steef Smit, Christine Au-Yeung, Berend Zwartsenberg, Maximilian Krautloher, Ilya S Elfimov, Roland Koch, Sergey Gorovikov, Chris Jozwiak, Aaron Bostwick, Eli Rotenberg, Bernhard Keimer, Andrea Damascelli

Abstract: The Mott insulator Ca$_2$RuO$_4$ exhibits a rare insulator-to-metal transition (IMT) induced by DC current. While structural changes associated with this transition have been tracked by neutron diffraction, Raman scattering, and x-ray spectroscopy, work on elucidating the response of the electronic degrees of freedom is still in progress. Here we unveil the current-induced modifications of the ele… ▽ More The Mott insulator Ca$_2$RuO$_4$ exhibits a rare insulator-to-metal transition (IMT) induced by DC current. While structural changes associated with this transition have been tracked by neutron diffraction, Raman scattering, and x-ray spectroscopy, work on elucidating the response of the electronic degrees of freedom is still in progress. Here we unveil the current-induced modifications of the electronic states of Ca$_2$RuO$_4$ by employing angle-resolved photoemission spectroscopy (ARPES) in conjunction with four-probe transport. Two main effects emerge: a clear reduction of the Mott gap and a modification in the dispersion of the Ru-bands. The changes in dispersion occur exclusively along the $XM$ high-symmetry direction, parallel to the $b$-axis where the greatest in-plane lattice change occurs. These experimental observations are reflected in dynamical mean-field theory (DMFT) calculations simulated exclusively from the current-induced lattice constants, indicating a current driven structural transition as the primary mechanism of the IMT. Furthermore, we demonstrate this phase is distinct from the high-temperature zero-current metallic phase. Our results provide insight into the elusive nature of the current-induced IMT of Ca$_2$RuO$_4$ and advance the challenging, yet powerful, technique of transport-ARPES. △ Less

Submitted 15 August, 2023; v1 submitted 10 August, 2023; originally announced August 2023.

Comments: 8 pages, 4 figures

Microscopic Entanglement Wedges

Authors: Robert de Mello Koch

Abstract: We study the holographic duality between the free $O(N)$ vector model and higher spin gravity. Conserved spinning primary currents of the conformal field theory (CFT) are dual to spinning gauge fields in the gravity. Reducing to independent components of the conserved CFT currents one finds two components at each spin. After gauge fixing the gravity and then reducing to independent components, one… ▽ More We study the holographic duality between the free $O(N)$ vector model and higher spin gravity. Conserved spinning primary currents of the conformal field theory (CFT) are dual to spinning gauge fields in the gravity. Reducing to independent components of the conserved CFT currents one finds two components at each spin. After gauge fixing the gravity and then reducing to independent components, one finds two components of the gauge field at each spin. Collective field theory provides a systematic way to map between these two sets of degrees of freedom, providing a complete and explicit identification between the dynamical degrees of freedom of the CFT and the dual gravity. The resulting map exhibits many features expected of holographic duality: it provides a valid bulk reconstruction, it reproduces insights expected from the holography of information and it provides a microscopic derivation of entanglement wedge reconstruction. △ Less

Submitted 14 July, 2023; v1 submitted 11 July, 2023; originally announced July 2023.

Comments: 1+45 pages; v2: minor typos corrected

Annotating Ambiguous Images: General Annotation Strategy for High-Quality Data with Real-World Biomedical Validation

Authors: Lars Schmarje, Vasco Grossmann, Claudius Zelenka, Johannes Brünger, Reinhard Koch

Abstract: In the field of image classification, existing methods often struggle with biased or ambiguous data, a prevalent issue in real-world scenarios. Current strategies, including semi-supervised learning and class blending, offer partial solutions but lack a definitive resolution. Addressing this gap, our paper introduces a novel strategy for generating high-quality labels in challenging datasets. Cent… ▽ More In the field of image classification, existing methods often struggle with biased or ambiguous data, a prevalent issue in real-world scenarios. Current strategies, including semi-supervised learning and class blending, offer partial solutions but lack a definitive resolution. Addressing this gap, our paper introduces a novel strategy for generating high-quality labels in challenging datasets. Central to our approach is a clearly designed flowchart, based on a broad literature review, which enables the creation of reliable labels. We validate our methodology through a rigorous real-world test case in the biomedical field, specifically in deducing height reduction from vertebral imaging. Our empirical study, leveraging over 250,000 annotations, demonstrates the effectiveness of our strategies decisions compared to their alternatives. △ Less

Submitted 29 April, 2024; v1 submitted 21 June, 2023; originally announced June 2023.

Comments: Accepted at ICLR 2024, DMLR Workshop

Label Smarter, Not Harder: CleverLabel for Faster Annotation of Ambiguous Image Classification with Higher Quality

Authors: Lars Schmarje, Vasco Grossmann, Tim Michels, Jakob Nazarenus, Monty Santarossa, Claudius Zelenka, Reinhard Koch

Abstract: High-quality data is crucial for the success of machine learning, but labeling large datasets is often a time-consuming and costly process. While semi-supervised learning can help mitigate the need for labeled data, label quality remains an open issue due to ambiguity and disagreement among annotators. Thus, we use proposal-guided annotations as one option which leads to more consistency between a… ▽ More High-quality data is crucial for the success of machine learning, but labeling large datasets is often a time-consuming and costly process. While semi-supervised learning can help mitigate the need for labeled data, label quality remains an open issue due to ambiguity and disagreement among annotators. Thus, we use proposal-guided annotations as one option which leads to more consistency between annotators. However, proposing a label increases the probability of the annotators deciding in favor of this specific label. This introduces a bias which we can simulate and remove. We propose a new method CleverLabel for Cost-effective LabEling using Validated proposal-guidEd annotations and Repaired LABELs. CleverLabel can reduce labeling costs by up to 30.0%, while achieving a relative improvement in Kullback-Leibler divergence of up to 29.8% compared to the previous state-of-the-art on a multi-domain real-world image classification benchmark. CleverLabel offers a novel solution to the challenge of efficiently labeling large datasets while also improving the label quality. △ Less

Submitted 22 May, 2023; originally announced May 2023.

Adversarial Hamiltonian learning of quantum dots in a minimal Kitaev chain

Authors: Rouven Koch, David van Driel, Alberto Bordin, Jose L. Lado, Eliska Greplova

Abstract: Determining Hamiltonian parameters from noisy experimental measurements is a key task for the control of experimental quantum systems. An experimental platform that recently emerged, and where knowledge of Hamiltonian parameters is crucial to fine-tune the system, is that of quantum dot-based Kitaev chains. In this work, we demonstrate an adversarial machine learning algorithm to determine the par… ▽ More Determining Hamiltonian parameters from noisy experimental measurements is a key task for the control of experimental quantum systems. An experimental platform that recently emerged, and where knowledge of Hamiltonian parameters is crucial to fine-tune the system, is that of quantum dot-based Kitaev chains. In this work, we demonstrate an adversarial machine learning algorithm to determine the parameters of a quantum dot-based Kitaev chain. We train a convolutional conditional generative adversarial neural network (Conv-cGAN) with simulated differential conductance data and use the model to predict the parameters at which Majorana bound states are predicted to appear. In particular, the Conv-cGAN model facilitates a rapid, numerically efficient exploration of the phase diagram describing the transition between elastic co-tunneling and crossed Andreev reflection regimes. We verify the theoretical predictions of the model by applying it to experimentally measured conductance obtained from a minimal Kitaev chain consisting of two spin-polarized quantum dots coupled by a superconductor-semiconductor hybrid. Our model accurately predicts, with an average success probability of $97$\%, whether the measurement was taken in the elastic co-tunneling or crossed Andreev reflection-dominated regime. Our work constitutes a step** stone towards fast, reliable parameter prediction for tuning quantum-dot systems into distinct Hamiltonian regimes. Ultimately, our results yield a strategy to support Kitaev chain tuning that is scalable to longer chains. △ Less

Submitted 21 April, 2023; originally announced April 2023.

Journal ref: Phys. Rev. Applied 20, 044081 (2023)

Exact Thermodynamics and Transport in the Classical Sine-Gordon Model

Authors: Rebekka Koch, Alvise Bastianello

Abstract: We revisit the exact thermodynamic description of the classical sine-Gordon field theory, a notorious integrable model. We found that existing results in the literature based on the soliton-gas picture did not correctly take into account light, but extended, solitons and thus led to incorrect results. This issue is regularized upon requantization: we derive the correct thermodynamics by taking the… ▽ More We revisit the exact thermodynamic description of the classical sine-Gordon field theory, a notorious integrable model. We found that existing results in the literature based on the soliton-gas picture did not correctly take into account light, but extended, solitons and thus led to incorrect results. This issue is regularized upon requantization: we derive the correct thermodynamics by taking the semiclassical limit of the quantum model. Our results are then extended to transport settings by means of Generalized Hydrodynamics. △ Less

Submitted 8 April, 2023; v1 submitted 29 March, 2023; originally announced March 2023.

Comments: 33 pages, 4 figures

Journal ref: SciPost Phys. 15, 140 (2023)

Temporal localized states and square-waves in semiconductor micro-resonators with strong time-delayed feedback

Authors: Elias R. Koch, Thomas G. Seidel, Julien Javaloyes, Svetlana V. Gurevich

Abstract: In this paper we study the dynamics of a vertically-emitting micro-cavity operated in the Gires-Tournois regime that contains a semiconductor quantum-well and that is subjected to strong time-delayed optical feedback and detuned optical injection. Using a first principle time-delay model for the optical response, we disclose sets of multistable dark and bright temporal localized states coexisting… ▽ More In this paper we study the dynamics of a vertically-emitting micro-cavity operated in the Gires-Tournois regime that contains a semiconductor quantum-well and that is subjected to strong time-delayed optical feedback and detuned optical injection. Using a first principle time-delay model for the optical response, we disclose sets of multistable dark and bright temporal localized states coexisting on their respective bistable homogeneous backgrounds. In the case of anti-resonant optical feedback, we disclose square-waves with a periodic of twice the round-trip in the external cavity. Finally, we perform a multiple time-scale analysis in the good cavity limit. The resulting normal form is in good agreement with the original time-delayed model. △ Less

Submitted 26 April, 2023; v1 submitted 23 January, 2023; originally announced January 2023.

Nanoscale view of engineered massive Dirac quasiparticles in lithographic superstructures

Authors: Alfred J. H. Jones, Lene Gammelgaard, Mikkel O. Sauer, Deepnarayan Biswas, Roland J. Koch, Chris Jozwiak, Eli Rotenberg, Aaron Bostwick, Kenji Watanabe, Takashi Taniguchi, Cory R. Dean, Antti-Pekka Jauho, Peter Bøggild, Thomas G. Pedersen, Bjarke S. Jessen, Søren Ulstrup

Abstract: Massive Dirac fermions are low-energy electronic excitations characterized by a hyperbolic band dispersion. They play a central role in several emerging physical phenomena such as topological phase transitions, anomalous Hall effects and superconductivity. This work demonstrates that massive Dirac fermions can be controllably induced by lithographically patterning superstructures of nanoscale hole… ▽ More Massive Dirac fermions are low-energy electronic excitations characterized by a hyperbolic band dispersion. They play a central role in several emerging physical phenomena such as topological phase transitions, anomalous Hall effects and superconductivity. This work demonstrates that massive Dirac fermions can be controllably induced by lithographically patterning superstructures of nanoscale holes in a graphene device. Their band dispersion is systematically visualized using angle-resolved photoemission spectroscopy with nanoscale spatial resolution. A linear scaling of effective mass with feature sizes is discovered, underlining the Dirac nature of the superstructures. In situ electrostatic do** dramatically enhances the effective hole mass and leads to the direct observation of an electronic band gap that results in a peak-to-peak band separation of (0.64 $\pm$ 0.03) eV, which is shown via first-principles calculations to be strongly renormalized by carrier-induced screening. The presented methodology outlines a new approach for band structure engineering guided by directly viewing structurally- and electrically-tunable massive Dirac quasiparticles in lithographic superstructures at the nanoscale. △ Less

Submitted 17 December, 2022; originally announced December 2022.

Comments: 37 pages, 12 figures (includes supporting information). A revised version has been published in ACS Nano

Journal ref: ACS Nano 2022, 16, 11, 19354-19362

Hamiltonian inference from dynamical excitations in confined quantum magnets

Authors: Netta Karjalainen, Zina Lippo, Guangze Chen, Rouven Koch, Adolfo O. Fumega, Jose L. Lado

Abstract: Quantum-disordered models provide a versatile platform to explore the emergence of quantum excitations in many-body systems. The engineering of spin models at the atomic scale with scanning tunneling microscopy and the local imaging of excitations with electrically driven spin resonance has risen as a powerful strategy to image spin excitations in finite quantum spin systems. Here, focusing on… ▽ More Quantum-disordered models provide a versatile platform to explore the emergence of quantum excitations in many-body systems. The engineering of spin models at the atomic scale with scanning tunneling microscopy and the local imaging of excitations with electrically driven spin resonance has risen as a powerful strategy to image spin excitations in finite quantum spin systems. Here, focusing on $S=1/2$ lattices as realized by Ti in MgO, we show that dynamical spin excitations provide a robust strategy to infer the nature of the underlying Hamiltonian. We show that finite-size interference of the dynamical many-body spin excitations of a generalized long-range Heisenberg model allows the underlying spin couplings to be inferred. We show that the spatial distribution of local spin excitations in Ti islands and ladders directly correlates with the underlying ground state in the thermodynamic limit. Using a supervised learning algorithm, we demonstrate that the different parameters of the Hamiltonian can be extracted by providing the spatially and frequency-dependent local excitations that can be directly measured by electrically driven spin resonance with scanning tunneling microscopy. Our results put forward local dynamical excitations in confined quantum spin models as versatile witnesses of the underlying ground state, providing an experimentally robust strategy for Hamiltonian inference in complex real spin models. △ Less

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

Comments: 11 pages, 10 figures

Journal ref: Phys. Rev. Applied 20, 024054 (2023)

Generating Functions for Giant Graviton Bound States

Authors: Warren Carlson, Robert de Mello Koch, Minkyoo Kim

Abstract: We construct generating functions for operators dual to systems of giant gravitons with open strings attached. These operators have a bare dimension of order $N$ so that the usual methods used to solve the planar limit are not applicable. The generating functions are given as integrals over auxiliary variables, which implement symmetrization and antisymmetrization of the indices of the fields from… ▽ More We construct generating functions for operators dual to systems of giant gravitons with open strings attached. These operators have a bare dimension of order $N$ so that the usual methods used to solve the planar limit are not applicable. The generating functions are given as integrals over auxiliary variables, which implement symmetrization and antisymmetrization of the indices of the fields from which the operator is composed. Operators of a good scaling dimension (eigenstates of the dilatation operator) are known as Gauss graph operators. Our generating functions give a simple construction of the Gauss graph operators which were previously obtained using a Fourier transform on a double coset. The new description provides a natural starting point for a systematic ${1\over N}$ expansion for these operators as well as the action of the dilatation operator on them, in terms of a saddle point evaluation of their integral representation. △ Less

Submitted 19 January, 2023; v1 submitted 13 December, 2022; originally announced December 2022.

Comments: 71+1 pages; Version 2 is JHEP version

$\mathcal{N}=4$ SYM, (super)-polynomial rings and emergent quantum mechanical symmetries

Authors: Robert de Mello Koch, Sanjaye Ramgoolam

Abstract: The structure of half-BPS representations of psu$(2,2|4)$ leads to the definition of a super-polynomial ring $\mathcal{R}(8|8)$ which admits a realisation of psu$(2,2|4)$ in terms of differential operators on the super-ring. The character of the half-BPS fundamental field representation encodes the resolution of the representation in terms of an exact sequence of modules of $\mathcal{R}(8|8)$. The… ▽ More The structure of half-BPS representations of psu$(2,2|4)$ leads to the definition of a super-polynomial ring $\mathcal{R}(8|8)$ which admits a realisation of psu$(2,2|4)$ in terms of differential operators on the super-ring. The character of the half-BPS fundamental field representation encodes the resolution of the representation in terms of an exact sequence of modules of $\mathcal{R}(8|8)$. The half-BPS representation is realized by quotienting the super-ring by a quadratic ideal, equivalently by setting to zero certain quadratic polynomials in the generators of the super-ring. This description of the half-BPS fundamental field irreducible representation of psu$(2,2|4)$ in terms of a super-polynomial ring is an example of a more general construction of lowest-weight representations of Lie (super-) algebras using polynomial rings generated by a commuting subspace of the standard raising operators, corresponding to positive roots of the Lie (super-) algebra. We illustrate the construction using simple examples of representations of su(3) and su(4). These results lead to the definition of a notion of quantum mechanical emergence for oscillator realisations of symmetries, which is based on ideals in the ring of polynomials in the creation operators. △ Less

Submitted 15 December, 2022; v1 submitted 8 November, 2022; originally announced November 2022.

Comments: 60 pages, no figures; v2: Mathematica notebook added to submission

Report number: QMUL-PH-22-34

Holography of Information in AdS/CFT

Authors: Robert de Mello Koch, Garreth Kemp

Abstract: The principle of the holography of information states that in a theory of quantum gravity a copy of all the information available on a Cauchy slice is also available near the boundary of the Cauchy slice. This redundancy in the theory is already present at low energy. In the context of the AdS/CFT correspondence, this principle can be translated into a statement about the dual conformal field theo… ▽ More The principle of the holography of information states that in a theory of quantum gravity a copy of all the information available on a Cauchy slice is also available near the boundary of the Cauchy slice. This redundancy in the theory is already present at low energy. In the context of the AdS/CFT correspondence, this principle can be translated into a statement about the dual conformal field theory. We carry out this translation and demonstrate that the principle of the holography of information holds in bilocal holography. △ Less

Submitted 20 October, 2022; originally announced October 2022.

Comments: 29+1 pages, 9 Figures

Non-local Skyrmions as topologically resilient quantum entangled states of light

Authors: Pedro Ornelas, Isaac Nape, Robert de Mello Koch, Andrew Forbes

Abstract: In the early 1960s, inspired by develo** notions of topological structure, Tony Skyrme suggested that sub-atomic particles be described as natural excitations of a single quantum field. Although never adopted for its intended purpose, the notion of a skyrmion as a topologically stable field configuration has proven highly versatile, finding application in condensed matter physics, acoustics and… ▽ More In the early 1960s, inspired by develo** notions of topological structure, Tony Skyrme suggested that sub-atomic particles be described as natural excitations of a single quantum field. Although never adopted for its intended purpose, the notion of a skyrmion as a topologically stable field configuration has proven highly versatile, finding application in condensed matter physics, acoustics and more recently optics, but all realised as localised fields and particles. Here we report the first non-local quantum entangled state with a non-trivial topology that is skyrmionic in nature, even though each individual photon has no salient topological structure. We demonstrate how the topology of the quantum wavefunction makes such quantum states robust to entanglement decay, remaining intact until the entanglement itself vanishes. Our work points to a nascent connection between entanglement classes and topology, holding exciting promise for the creation and preservation of quantum information by topologically structured quantum states that persist even when entanglement is fragile. △ Less

Submitted 15 March, 2023; v1 submitted 10 October, 2022; originally announced October 2022.

Role of Local Ru Hexamers in Superconductivity of Ruthenium Phosphide

Authors: Robert J. Koch, Niraj Aryal, Oleh Ivashko, Yu Liu, Milinda Abeykoon, Eric D. Bauer, Martin v. Zimmermann, Weiguo Yin, Cedomir Petrovic, Emil S. Bozin

Abstract: Superconductivity in binary ruthenium pnictides occurs proximal to and upon suppression of a mysterious non-magnetic ground state, preceded by a pseudogap phase associated with Fermi surface instability, and its critical temperature, T$_{c}$, is maximized around the pseudogap quantum critical point. By analogy with isoelectronic iron based counterparts, antiferromagnetic fluctuations became "usual… ▽ More Superconductivity in binary ruthenium pnictides occurs proximal to and upon suppression of a mysterious non-magnetic ground state, preceded by a pseudogap phase associated with Fermi surface instability, and its critical temperature, T$_{c}$, is maximized around the pseudogap quantum critical point. By analogy with isoelectronic iron based counterparts, antiferromagnetic fluctuations became "usual suspects" as putative mediators of superconducting pairing. Here we report on a high temperature local symmetry breaking in RuP, the parent of the maximum-Tc branch of these novel superconductors, revealed by combined nanostructure-sensitive powder and single crystal X-ray total scattering experiments. Large local Ru$_{6}$ hexamer distortions associated with orbital-charge trimerization form above the two-stage electronic transition in RuP. While hexamer ordering enables the nonmagnetic ground state and presumed complex oligomerization, the relevance of pseudogap fluctuations for superconductivity emerges as a distinct prospect. As a transition metal system in which partial d-manifold filling combined with high crystal symmetry promotes electronic instabilities, this represents a further example of local electronic precursors underpinning the macroscopic collective behavior of quantum materials. △ Less

Submitted 19 August, 2022; originally announced August 2022.

Comments: 10 pages 6 figures

Journal ref: Phys. Rev. B 106, 214516 (2022)

Automated Classification of Nanoparticles with Various Ultrastructures and Sizes

Authors: Claudius Zelenka, Marius Kamp, Kolja Strohm, Akram Kadoura, Jacob Johny, Reinhard Koch, Lorenz Kienle

Abstract: Accurately measuring the size, morphology, and structure of nanoparticles is very important, because they are strongly dependent on their properties for many applications. In this paper, we present a deep-learning based method for nanoparticle measurement and classification trained from a small data set of scanning transmission electron microscopy images. Our approach is comprised of two stages: l… ▽ More Accurately measuring the size, morphology, and structure of nanoparticles is very important, because they are strongly dependent on their properties for many applications. In this paper, we present a deep-learning based method for nanoparticle measurement and classification trained from a small data set of scanning transmission electron microscopy images. Our approach is comprised of two stages: localization, i.e., detection of nanoparticles, and classification, i.e., categorization of their ultrastructure. For each stage, we optimize the segmentation and classification by analysis of the different state-of-the-art neural networks. We show how the generation of synthetic images, either using image processing or using various image generation neural networks, can be used to improve the results in both stages. Finally, the application of the algorithm to bimetallic nanoparticles demonstrates the automated data collection of size distributions including classification of complex ultrastructures. The developed method can be easily transferred to other material systems and nanoparticle structures. △ Less

Submitted 28 July, 2022; originally announced July 2022.

Opportunistic hip fracture risk prediction in Men from X-ray: Findings from the Osteoporosis in Men (MrOS) Study

Authors: Lars Schmarje, Stefan Reinhold, Timo Damm, Eric Orwoll, Claus-C. Glüer, Reinhard Koch

Abstract: Osteoporosis is a common disease that increases fracture risk. Hip fractures, especially in elderly people, lead to increased morbidity, decreased quality of life and increased mortality. Being a silent disease before fracture, osteoporosis often remains undiagnosed and untreated. Areal bone mineral density (aBMD) assessed by dual-energy X-ray absorptiometry (DXA) is the gold-standard method for o… ▽ More Osteoporosis is a common disease that increases fracture risk. Hip fractures, especially in elderly people, lead to increased morbidity, decreased quality of life and increased mortality. Being a silent disease before fracture, osteoporosis often remains undiagnosed and untreated. Areal bone mineral density (aBMD) assessed by dual-energy X-ray absorptiometry (DXA) is the gold-standard method for osteoporosis diagnosis and hence also for future fracture prediction (prognostic). However, the required special equipment is not broadly available everywhere, in particular not to patients in develo** countries. We propose a deep learning classification model (FORM) that can directly predict hip fracture risk from either plain radiographs (X-ray) or 2D projection images of computed tomography (CT) data. Our method is fully automated and therefore well suited for opportunistic screening settings, identifying high risk patients in a broader population without additional screening. FORM was trained and evaluated on X-rays and CT projections from the Osteoporosis in Men (MrOS) study. 3108 X-rays (89 incident hip fractures) or 2150 CTs (80 incident hip fractures) with a 80/20 split were used. We show that FORM can correctly predict the 10-year hip fracture risk with a validation AUC of 81.44 +- 3.11% / 81.04 +- 5.54% (mean +- STD) including additional information like age, BMI, fall history and health background across a 5-fold cross validation on the X-ray and CT cohort, respectively. Our approach significantly (p < 0.01) outperforms previous methods like Cox Proportional-Hazards Model and \frax with 70.19 +- 6.58 and 74.72 +- 7.21 respectively on the X-ray cohort. Our model outperform on both cohorts hip aBMD based predictions. We are confident that FORM can contribute on improving osteoporosis diagnosis at an early stage. △ Less

Submitted 6 October, 2022; v1 submitted 22 July, 2022; originally announced July 2022.

Comments: Oral Presentation at MICCAI 2022 Workshop (PRIME), Considered for best paper award Predictive Intelligence in Medicine. PRIME 2022. Lecture Notes in Computer Science, vol 13564

Beyond Hard Labels: Investigating data label distributions

Authors: Vasco Grossmann, Lars Schmarje, Reinhard Koch

Abstract: High-quality data is a key aspect of modern machine learning. However, labels generated by humans suffer from issues like label noise and class ambiguities. We raise the question of whether hard labels are sufficient to represent the underlying ground truth distribution in the presence of these inherent imprecision. Therefore, we compare the disparity of learning with hard and soft labels quantita… ▽ More High-quality data is a key aspect of modern machine learning. However, labels generated by humans suffer from issues like label noise and class ambiguities. We raise the question of whether hard labels are sufficient to represent the underlying ground truth distribution in the presence of these inherent imprecision. Therefore, we compare the disparity of learning with hard and soft labels quantitatively and qualitatively for a synthetic and a real-world dataset. We show that the application of soft labels leads to improved performance and yields a more regular structure of the internal feature space. △ Less

Submitted 6 October, 2022; v1 submitted 13 July, 2022; originally announced July 2022.

Comments: https://icml.cc/virtual/2022/workshop/13477

Journal ref: ICML 2022 Workshop DataPerf: Benchmarking Data for Data-Centric AI

Is one annotation enough? A data-centric image classification benchmark for noisy and ambiguous label estimation

Authors: Lars Schmarje, Vasco Grossmann, Claudius Zelenka, Sabine Dippel, Rainer Kiko, Mariusz Oszust, Matti Pastell, Jenny Stracke, Anna Valros, Nina Volkmann, Reinhard Koch

Abstract: High-quality data is necessary for modern machine learning. However, the acquisition of such data is difficult due to noisy and ambiguous annotations of humans. The aggregation of such annotations to determine the label of an image leads to a lower data quality. We propose a data-centric image classification benchmark with ten real-world datasets and multiple annotations per image to allow researc… ▽ More High-quality data is necessary for modern machine learning. However, the acquisition of such data is difficult due to noisy and ambiguous annotations of humans. The aggregation of such annotations to determine the label of an image leads to a lower data quality. We propose a data-centric image classification benchmark with ten real-world datasets and multiple annotations per image to allow researchers to investigate and quantify the impact of such data quality issues. With the benchmark we can study the impact of annotation costs and (semi-)supervised methods on the data quality for image classification by applying a novel methodology to a range of different algorithms and diverse datasets. Our benchmark uses a two-phase approach via a data label improvement method in the first phase and a fixed evaluation model in the second phase. Thereby, we give a measure for the relation between the input labeling effort and the performance of (semi-)supervised algorithms to enable a deeper insight into how labels should be created for effective model training. Across thousands of experiments, we show that one annotation is not enough and that the inclusion of multiple annotations allows for a better approximation of the real underlying class distribution. We identify that hard labels can not capture the ambiguity of the data and this might lead to the common issue of overconfident models. Based on the presented datasets, benchmarked methods, and analysis, we create multiple research opportunities for the future directed at the improvement of label noise estimation approaches, data annotation schemes, realistic (semi-)supervised learning, or more reliable image collection. △ Less

Submitted 4 November, 2022; v1 submitted 13 July, 2022; originally announced July 2022.

Comments: Accepted at NeurIPS 2022, Benchmark and Dataset Track, Code and Link to data available at https://github.com/Emprime/dcic

Smoothed Particle Hydrodynamics Physically Reconsidered -- The Relation to Explicit Large Eddy Simulation and the Issue of Particle Duality

Authors: Max Okraschevski, Niklas Bürkle, Rainer Koch, Hans-Jörg Bauer

Abstract: In this work we will identify a novel relation between Smoothed Particle Hydrodynamics (SPH) and explicit Large Eddy Simulation (LES) using a coarse-graining method from Non-Equilibrium Molecular Dynamics (NEMD). While the current literature points at the conclusion that characteristic SPH issues become restrictive for subsonic turbulent flows, we see the potential to mitigate these SPH issues by… ▽ More In this work we will identify a novel relation between Smoothed Particle Hydrodynamics (SPH) and explicit Large Eddy Simulation (LES) using a coarse-graining method from Non-Equilibrium Molecular Dynamics (NEMD). While the current literature points at the conclusion that characteristic SPH issues become restrictive for subsonic turbulent flows, we see the potential to mitigate these SPH issues by explicit subfilter stress (SFS) modelling. We verify our theory by various simulations of homogeneous, isotropic turbulence (HIT) at $Re=10^4$ and compare the results to a Direct Numerical Simulation (DNS) reported by Dairay et al. (2017). Although the simulations substantiate our theory, we see another issue arising, which is conceptually rooted in the particle itself, termed as Particle Duality. Finally, we conclude our work by acknowledging SPH as coarse-graining method for turbulent flows, highlighting its capabilities and limitations. △ Less

Submitted 22 November, 2022; v1 submitted 21 June, 2022; originally announced June 2022.

Comments: Added Journal Reference & DOI

Journal ref: Physics of Fluids 34, 115108 (2022)

Square wave generation in vertical external-cavity Kerr-Gires-Tournois interferometers

Authors: Elias R. Koch, Thomas G. Seidel, Svetlana V. Gurevich, Julien Javaloyes

Abstract: We study theoretically the mechanisms of square-wave (SW) formation in vertical external-cavity Kerr-Gires-Tournois interferometers in presence of anti-resonant injection. We provide simple analytical approximations for their plateau intensities and for the conditions of their emergence. We demonstrate that SWs may appear via a homoclinic snaking scenario, leading to the formation of complex-shape… ▽ More We study theoretically the mechanisms of square-wave (SW) formation in vertical external-cavity Kerr-Gires-Tournois interferometers in presence of anti-resonant injection. We provide simple analytical approximations for their plateau intensities and for the conditions of their emergence. We demonstrate that SWs may appear via a homoclinic snaking scenario, leading to the formation of complex-shaped multistable SW solutions. The resulting SWs can host localized structures and robust bound-states. △ Less

Submitted 28 July, 2022; v1 submitted 20 June, 2022; originally announced June 2022.

Comments: 4 pages, 5 figures, submitted to optics letters

Large-Area Intercalated 2D-Pb/Graphene Heterostructure as a Platform for Generating Spin-Orbit Torque

Authors: Alexander Vera, Boyang Zheng, Wilson Yanez, Kaijie Yang, Seong Yeoul Kim, Jimmy C. Kotsakidis, Hesham El-Sherif, Gopi Krishnan, Roland J. Koch, Timothy A. Bowen, Chengye Dong, Yuanxi Wang, Maxwell Wetherington, Eli Rotenberg, Nabil Bassim, Adam L. Friedman, Robert M. Wallace, Chaoxing Liu, Nitin Samarth, Vincent H. Crespi, Joshua A. Robinson

Abstract: A scalable platform to synthesize ultrathin heavy metals may enable high efficiency charge-to-spin conversion for next-generation spintronics. Here we report centimeter-scale synthesis of air-stable, epitaxially registered monolayer Pb underneath bilayer graphene on SiC (0001) by confinement heteroepitaxy (CHet). Diffraction, spectroscopy, and microscopy reveal CHet-based Pb intercalation predomin… ▽ More A scalable platform to synthesize ultrathin heavy metals may enable high efficiency charge-to-spin conversion for next-generation spintronics. Here we report centimeter-scale synthesis of air-stable, epitaxially registered monolayer Pb underneath bilayer graphene on SiC (0001) by confinement heteroepitaxy (CHet). Diffraction, spectroscopy, and microscopy reveal CHet-based Pb intercalation predominantly exhibits a mottled hexagonal superstructure due to an ordered network of Frenkel-Kontorova-like domain walls. The system's air stability enables ex-situ spin torque ferromagnetic resonance (ST-FMR) measurements that demonstrate charge-to-spin conversion in graphene/Pb/ferromagnet heterostructures with a 1.5x increase in the effective field ratio compared to control samples. △ Less

Submitted 27 March, 2024; v1 submitted 13 May, 2022; originally announced May 2022.

Comments: 27 pages, 5 figures. Supporting Information included (16 pages, 14 figures, 1 table)

Strongly-overdoped La$_{2-x}$Sr$_x$CuO$_4$: Evidence for Josephson-coupled grains of strongly-correlated superconductor

Authors: Yangmu Li, A. Sapkota, P. M. Lozano, Zengyi Du, Hui Li, Zebin Wu, Asish K. Kundu, R. J. Koch, Lijun Wu, B. L. Winn, Songxue Chi, M. Matsuda, M. Frontzek, E. S. Bozin, Yimei Zhu, I. Bozovic, Abhay N. Pasupathy, Ilya K. Drozdov, Kazuhiro Fujita, G. D. Gu, Igor Zaliznyak, Qiang Li, J. M. Tranquada

Abstract: The interpretation of how superconductivity disappears in cuprates at large hole do** has been controversial. To address this issue, we present an experimental study of single-crystal and thin film samples of La$_{2-x}$Sr$_x$CuO$_4$ (LSCO) with $x\ge0.25$. In particular, measurements of bulk susceptibility on LSCO crystals with $x=0.25$ indicate an onset of diamagnetism at $T_{c1}=38.5$ K, with… ▽ More The interpretation of how superconductivity disappears in cuprates at large hole do** has been controversial. To address this issue, we present an experimental study of single-crystal and thin film samples of La$_{2-x}$Sr$_x$CuO$_4$ (LSCO) with $x\ge0.25$. In particular, measurements of bulk susceptibility on LSCO crystals with $x=0.25$ indicate an onset of diamagnetism at $T_{c1}=38.5$ K, with a sharp transition to a phase with full bulk shielding at $T_{c2}=18$ K, independent of field direction. Strikingly, the in-plane resistivity only goes to zero at $T_{c2}$. Inelastic neutron scattering on $x=0.25$ crystals confirms the presence of low-energy incommensurate magnetic excitations with reduced strength compared to lower do** levels. The ratio of the spin gap to $T_{c2}$ is anomalously large. Our results are consistent with a theoretical prediction for strongly overdoped cuprates by Spivak, Oreto, and Kivelson, in which superconductivity initially develops within disconnected self-organized grains characterized by a reduced hole concentration, with bulk superconductivity occurring only after superconductivity is induced by proximity effect in the surrounding medium of higher hole concentration. Beyond the superconducting-to-metal transition, local differential conductance measurements on an LSCO thin film suggest that regions with pairing correlations survive, but are too dilute to support superconducting order. Future experiments will be needed to test the degree to which these results apply to overdoped cuprates in general. △ Less

Submitted 21 December, 2022; v1 submitted 3 May, 2022; originally announced May 2022.

Comments: 21 pages, 15 figures; added more sample characterizations, results for another sample, more discussion; accepted version

Journal ref: Phys. Rev. B 106, 224515 (2022)

Modern applications of machine learning in quantum sciences

Authors: Anna Dawid, Julian Arnold, Borja Requena, Alexander Gresch, Marcin Płodzień, Kaelan Donatella, Kim A. Nicoli, Paolo Stornati, Rouven Koch, Miriam Büttner, Robert Okuła, Gorka Muñoz-Gil, Rodrigo A. Vargas-Hernández, Alba Cervera-Lierta, Juan Carrasquilla, Vedran Dunjko, Marylou Gabrié, Patrick Huembeli, Evert van Nieuwenburg, Filippo Vicentini, Lei Wang, Sebastian J. Wetzel, Giuseppe Carleo, Eliška Greplová, Roman Krems , et al. (4 additional authors not shown)

Abstract: In this book, we provide a comprehensive introduction to the most recent advances in the application of machine learning methods in quantum sciences. We cover the use of deep learning and kernel methods in supervised, unsupervised, and reinforcement learning algorithms for phase classification, representation of many-body quantum states, quantum feedback control, and quantum circuits optimization.… ▽ More In this book, we provide a comprehensive introduction to the most recent advances in the application of machine learning methods in quantum sciences. We cover the use of deep learning and kernel methods in supervised, unsupervised, and reinforcement learning algorithms for phase classification, representation of many-body quantum states, quantum feedback control, and quantum circuits optimization. Moreover, we introduce and discuss more specialized topics such as differentiable programming, generative models, statistical approach to machine learning, and quantum machine learning. △ Less

Submitted 15 November, 2023; v1 submitted 8 April, 2022; originally announced April 2022.

Comments: 288 pages, 92 figures. We have a publishing contract with Cambridge University Press. Figures and tex files are available at https://github.com/Shmoo137/Lecture-Notes

Machine Learning Integrated with Model Predictive Control for Imitative Optimal Control of Compression Ignition Engines

Authors: Armin Norouzi, Saeid Shahpouri, David Gordon, Alexander Winkler, Eugen Nuss, Dirk Abel, Jakob Andert, Mahdi Shahbakhti, Charles Robert Koch

Abstract: The high thermal efficiency and reliability of the compression-ignition engine makes it the first choice for many applications. For this to continue, a reduction of the pollutant emissions is needed. One solution is the use of machine learning (ML) and model predictive control (MPC) to minimize emissions and fuel consumption, without adding substantial computational cost to the engine controller.… ▽ More The high thermal efficiency and reliability of the compression-ignition engine makes it the first choice for many applications. For this to continue, a reduction of the pollutant emissions is needed. One solution is the use of machine learning (ML) and model predictive control (MPC) to minimize emissions and fuel consumption, without adding substantial computational cost to the engine controller. ML is developed in this paper for both modeling engine performance and emissions and for imitating the behaviour of an Linear Parameter Varying (LPV) MPC. Using a support vector machine-based linear parameter varying model of the engine performance and emissions, a model predictive controller is implemented for a 4.5 Cummins diesel engine. This online optimized MPC solution offers advantages in minimizing the \nox~emissions and fuel consumption compared to the baseline feedforward production controller. To reduce the computational cost of this MPC, a deep learning scheme is designed to mimic the behavior of the developed controller. The performance in reducing NOx emissions at a constant load by the imitative controller is similar to that of the online optimized MPC compared to the Cummins production controller. In addition, the imitative controller requires 50 times less computation time compared to that of the online MPC optimization. △ Less

Submitted 2 August, 2022; v1 submitted 31 March, 2022; originally announced April 2022.

Comments: Submitted to Advances in Automotive Control - 10th AAC 2022

Deep Learning based Model Predictive Control for Compression Ignition Engines

Authors: Armin Norouzi, Saeid Shahpouri, David Gordon, Alexander Winkler, Eugen Nuss, Dirk Abel, Jakob Andert, Mahdi Shahbakhti, Charles Robert Koch

Abstract: Machine learning (ML) and a nonlinear model predictive controller (NMPC) are used in this paper to minimize the emissions and fuel consumption of a compression ignition engine. In this work machine learning is applied in two methods. In the first application, ML is used to identify a model for implementation in model predictive control optimization problems. In the second application, ML is used a… ▽ More Machine learning (ML) and a nonlinear model predictive controller (NMPC) are used in this paper to minimize the emissions and fuel consumption of a compression ignition engine. In this work machine learning is applied in two methods. In the first application, ML is used to identify a model for implementation in model predictive control optimization problems. In the second application, ML is used as a replacement of the NMPC where the ML controller learns the optimal control action by imitating or mimicking the behavior of the model predictive controller. In this study, a deep recurrent neural network including long-short term memory (LSTM) layers are used to model the emissions and performance of an industrial 4.5 liter 4-cylinder Cummins diesel engine. This model is then used for model predictive controller implementation. Then, a deep learning scheme is deployed to clone the behavior of the developed controller. In the LSTM integration, a novel scheme is used by augmenting hidden and cell states of the network in an NMPC optimization problem. The developed LSTM-NMPC and the imitative NMPC are compared with the Cummins calibrated Engine Control Unit (ECU) model in an experimentally validated engine simulation platform. Results show a significant reduction in Nitrogen Oxides (\nox) emissions and a slight decrease in the injected fuel quantity while maintaining the same load. In addition, the imitative NMPC has a similar performance as the NMPC but with a two orders of magnitude reduction of the computation time. △ Less

Submitted 4 August, 2022; v1 submitted 31 March, 2022; originally announced April 2022.

Comments: Submitted to Control engineering Practice (Submission date: March 9, 2022) Revised version (Submission date: June 18, 2022) Accepted on July 30, 2022

Simulation of Plenoptic Cameras

Authors: Tim Michels, Arne Petersen, Luca Palmieri, Reinhard Koch

Abstract: Plenoptic cameras enable the capturing of spatial as well as angular color information which can be used for various applications among which are image refocusing and depth calculations. However, these cameras are expensive and research in this area currently lacks data for ground truth comparisons. In this work we describe a flexible, easy-to-use Blender model for the different plenoptic camera t… ▽ More Plenoptic cameras enable the capturing of spatial as well as angular color information which can be used for various applications among which are image refocusing and depth calculations. However, these cameras are expensive and research in this area currently lacks data for ground truth comparisons. In this work we describe a flexible, easy-to-use Blender model for the different plenoptic camera types which is on the one hand able to provide the ground truth data for research and on the other hand allows an inexpensive assessment of the cameras usefulness for the desired applications. Furthermore we show that the rendering results exhibit the same image degradation effects as real cameras and make our simulation publicly available. △ Less

Submitted 9 March, 2022; originally announced March 2022.

Journal ref: 2018-3DTV-Conference: The True Vision-Capture, Transmission and Display of 3D Video (3DTV-CON). IEEE, 2018

Creating Realistic Ground Truth Data for the Evaluation of Calibration Methods for Plenoptic and Conventional Cameras

Authors: Tim Michels, Arne Petersen, Reinhard Koch

Abstract: Camera calibration methods usually consist of capturing images of known calibration patterns and using the detected correspondences to optimize the parameters of the assumed camera model. A meaningful evaluation of these methods relies on the availability of realistic synthetic data. In previous works concerned with conventional cameras the synthetic data was mainly created by rendering perfect im… ▽ More Camera calibration methods usually consist of capturing images of known calibration patterns and using the detected correspondences to optimize the parameters of the assumed camera model. A meaningful evaluation of these methods relies on the availability of realistic synthetic data. In previous works concerned with conventional cameras the synthetic data was mainly created by rendering perfect images with a pinhole camera and subsequently adding distortions and aberrations to the renderings and correspondences according to the assumed camera model. This method can bias the evaluation since not every camera perfectly complies with an assumed model. Furthermore, in the field of plenoptic camera calibration there is no synthetic ground truth data available at all. We address these problems by proposing a method based on backward ray tracing to create realistic ground truth data that can be used for an unbiased evaluation of calibration methods for both types of cameras. △ Less

Submitted 9 March, 2022; originally announced March 2022.

Comments: 9 pages, 8 figures. Accepted at 3DV 2019

Journal ref: 2019 International Conference on 3D Vision (3DV). IEEE, 2019

Ray Tracing-Guided Design of Plenoptic Cameras

Authors: Tim Michels, Reinhard Koch

Abstract: The design of a plenoptic camera requires the combination of two dissimilar optical systems, namely a main lens and an array of microlenses. And while the construction process of a conventional camera is mainly concerned with focusing the image onto a single plane, in the case of plenoptic cameras there can be additional requirements such as a predefined depth of field or a desired range of dispar… ▽ More The design of a plenoptic camera requires the combination of two dissimilar optical systems, namely a main lens and an array of microlenses. And while the construction process of a conventional camera is mainly concerned with focusing the image onto a single plane, in the case of plenoptic cameras there can be additional requirements such as a predefined depth of field or a desired range of disparities in neighboring microlens images. Due to this complexity, the manual creation of multiple plenoptic camera setups is often a time-consuming task. In this work we assume a simulation framework as well as the main lens data given and present a method to calculate the remaining aperture, sensor and microlens array parameters under different sets of constraints. Our ray tracing-based approach is shown to result in models outperforming their pendants generated with the commonly used paraxial approximations in terms of image quality, while still meeting the desired constraints. Both the implementation and evaluation setup including 30 plenoptic camera designs are made publicly available. △ Less

Submitted 9 March, 2022; originally announced March 2022.

Comments: 9 pages, 9 figures. Accepted at 3DV 2021. 2021 International Conference on 3D Vision (3DV). IEEE, 2021

Higher dimensional CFTs as 2D conformally-equivariant topological field theories

Authors: Robert de Mello Koch, Sanjaye Ramgoolam

Abstract: Two and three-point functions of primary fields in four dimensional CFT have a simple space-time dependences factored out from the combinatoric structure which enumerates the fields and gives their couplings. This has led to the formulation of two dimensional topological field theories with $so(4,2)$ equivariance which are conjectured to be equivalent to higher dimensional conformal field theories… ▽ More Two and three-point functions of primary fields in four dimensional CFT have a simple space-time dependences factored out from the combinatoric structure which enumerates the fields and gives their couplings. This has led to the formulation of two dimensional topological field theories with $so(4,2)$ equivariance which are conjectured to be equivalent to higher dimensional conformal field theories. We review this CFT4/TFT2 construction in the simplest possible setting of a free scalar field, which gives an algebraic construction of the correlators in terms of an infinite dimensional $so(4,2)$ equivariant algebra with finite dimensional subspaces at fixed scaling dimension. Crossing symmetry of the CFT4 is related to associativity of the algebra. This construction is then extended to describe perturbative CFT4, by making use of deformed co-products. Motivated by the Wilson-Fisher CFT we outline the construction of U(so($d$,2)) equivariant TFT2 for non-integer $d$, in terms of diagram algebras and their representations. △ Less

Submitted 18 February, 2022; originally announced February 2022.

Comments: 19 pages, written version of invited talk given by SR at the XIV International Workshop LIE THEORY AND ITS APPLICATIONS IN PHYSICS, Sofia

Report number: QMUL-PH-22-06

Designing quantum many-body matter with conditional generative adversarial networks

Authors: Rouven Koch, Jose L. Lado

Abstract: The computation of dynamical correlators of quantum many-body systems represents an open critical challenge in condensed matter physics. While powerful methodologies have risen in recent years, covering the full parameter space remains unfeasible for most many-body systems with a complex configuration space. Here we demonstrate that conditional Generative Adversarial Networks (GANs) allow simulati… ▽ More The computation of dynamical correlators of quantum many-body systems represents an open critical challenge in condensed matter physics. While powerful methodologies have risen in recent years, covering the full parameter space remains unfeasible for most many-body systems with a complex configuration space. Here we demonstrate that conditional Generative Adversarial Networks (GANs) allow simulating the full parameter space of several many-body systems, accounting both for controlled parameters, and stochastic disorder effects. After training with a restricted set of noisy many-body calculations, the conditional GAN algorithm provides the whole dynamical excitation spectra for a Hamiltonian instantly and with an accuracy analogous to the exact calculation. We further demonstrate how the trained conditional GAN automatically provides a powerful method for Hamiltonian learning from its dynamical excitations, and to flag non-physical systems via outlier detection. Our methodology puts forward generative adversarial learning as a powerful technique to explore complex many-body phenomena, providing a starting point to design large-scale quantum many-body matter. △ Less

Submitted 26 September, 2022; v1 submitted 28 January, 2022; originally announced January 2022.

Journal ref: Phys. Rev. Research 4, 033223 (2022)

Magnetism in Metastable and Annealed Compositionally Complex Alloys

Authors: Nan Tang, Lizabeth Quigley, Walker L. Boldman, Cameron S. Jorgensen, Rémi Koch, Daniel O'Leary, Hugh R. Meda, Philip D. Rack, Dustin A. Gilbert

Abstract: Compositionally complex materials (CCMs) present a potential paradigm shift in the design of magnetic materials. These alloys exhibit long-range structural order coupled with limited or no chemical order. As a result, extreme local environments exist with a large opposing magnetic energy term, which can manifest large changes in the magnetic behavior. In the current work, the magnetic properties o… ▽ More Compositionally complex materials (CCMs) present a potential paradigm shift in the design of magnetic materials. These alloys exhibit long-range structural order coupled with limited or no chemical order. As a result, extreme local environments exist with a large opposing magnetic energy term, which can manifest large changes in the magnetic behavior. In the current work, the magnetic properties of (Cr, Mn, Fe, Ni) alloys are presented. These materials were prepared by room-temperature combinatorial sputtering, resulting in a range of compositions with a single BCC structural phase and no chemical ordering. The combinatorial growth technique allows CCMs to be prepared outside of their thermodynamically stable phase, enabling the exploration of otherwise inaccessible order. The mixed ferromagnetic and antiferromagnetic interactions in these alloys causes frustrated magnetic behavior, which results in an extremely low coercivity (<1 mT), which increases rapidly at 50 K. At low temperatures, the coercivity achieves values of nearly 500 mT, which is comparable to some high-anisotropy magnetic materials. Commensurate with the divergent coercivity is an atypical drop in the temperature dependent magnetization. These effects are explained by a mixed magnetic phase model, consisting of ferro-, antiferro , and frustrated magnetic regions, and are rationalized by simulations. A machine-learning algorithm is employed to visualize the parameter space and inform the development of subsequent compositions. Annealing the samples at 600 °C orders the sample, more-than doubling the Curie temperature and increasing the saturation magnetization by as much as 5x. Simultaneously, the large coercivities are suppressed, resulting in magnetic behavior that is largely temperature independent over a range of 350 K. △ Less

Submitted 23 November, 2021; originally announced November 2021.

Comments: 13 pages, 6 figures

Journal ref: Phys. Rev. Materials 5, 114405 (2021)

Generalized Hydrodynamics of the attractive Non-Linear Schroedinger Equation

Authors: Rebekka Koch, Jean-Sébastien Caux, Alvise Bastianello

Abstract: We study the generalized hydrodynamics of the one-dimensional classical Non Linear Schroedinger equation in the attractive phase. We thereby show that the thermodynamic limit is entirely captured by solitonic modes and radiation is absent. Our results are derived by considering the semiclassical limit of the quantum Bose gas, where the Planck constant has a key role as a regulator of the classical… ▽ More We study the generalized hydrodynamics of the one-dimensional classical Non Linear Schroedinger equation in the attractive phase. We thereby show that the thermodynamic limit is entirely captured by solitonic modes and radiation is absent. Our results are derived by considering the semiclassical limit of the quantum Bose gas, where the Planck constant has a key role as a regulator of the classical soliton gas. We use our result to study adiabatic interaction changes from the repulsive to the attractive phase, observing soliton production and obtaining exact analytical results which are in excellent agreement with Monte Carlo simulations. △ Less

Submitted 27 October, 2021; originally announced October 2021.

Comments: 24 pages, 4 figures

Fuzzy Overclustering: Semi-Supervised Classification of Fuzzy Labels with Overclustering and Inverse Cross-Entropy

Authors: Lars Schmarje, Johannes Brünger, Monty Santarossa, Simon-Martin Schröder, Rainer Kiko, Reinhard Koch

Abstract: Deep learning has been successfully applied to many classification problems including underwater challenges. However, a long-standing issue with deep learning is the need for large and consistently labeled datasets. Although current approaches in semi-supervised learning can decrease the required amount of annotated data by a factor of 10 or even more, this line of research still uses distinct cla… ▽ More Deep learning has been successfully applied to many classification problems including underwater challenges. However, a long-standing issue with deep learning is the need for large and consistently labeled datasets. Although current approaches in semi-supervised learning can decrease the required amount of annotated data by a factor of 10 or even more, this line of research still uses distinct classes. For underwater classification, and uncurated real-world datasets in general, clean class boundaries can often not be given due to a limited information content in the images and transitional stages of the depicted objects. This leads to different experts having different opinions and thus producing fuzzy labels which could also be considered ambiguous or divergent. We propose a novel framework for handling semi-supervised classifications of such fuzzy labels. It is based on the idea of overclustering to detect substructures in these fuzzy labels. We propose a novel loss to improve the overclustering capability of our framework and show the benefit of overclustering for fuzzy labels. We show that our framework is superior to previous state-of-the-art semi-supervised methods when applied to real-world plankton data with fuzzy labels. Moreover, we acquire 5 to 10\% more consistent predictions of substructures. △ Less

Submitted 13 October, 2021; originally announced October 2021.

Comments: Source code: https://github.com/Emprime/FuzzyOverclustering Datasets: https://doi.org/10.5281/zenodo.5550918. arXiv admin note: substantial text overlap with arXiv:2012.01768

Journal ref: Sensors 2021, 21(19), 6661

Life is not black and white -- Combining Semi-Supervised Learning with fuzzy labels

Authors: Lars Schmarje, Reinhard Koch

Abstract: The required amount of labeled data is one of the biggest issues in deep learning. Semi-Supervised Learning can potentially solve this issue by using additional unlabeled data. However, many datasets suffer from variability in the annotations. The aggregated labels from these annotation are not consistent between different annotators and thus are considered fuzzy. These fuzzy labels are often not… ▽ More The required amount of labeled data is one of the biggest issues in deep learning. Semi-Supervised Learning can potentially solve this issue by using additional unlabeled data. However, many datasets suffer from variability in the annotations. The aggregated labels from these annotation are not consistent between different annotators and thus are considered fuzzy. These fuzzy labels are often not considered by Semi-Supervised Learning. This leads either to an inferior performance or to higher initial annotation costs in the complete machine learning development cycle. We envision the incorporation of fuzzy labels into Semi-Supervised Learning and give a proof-of-concept of the potential lower costs and higher consistency in the complete development cycle. As part of our concept, we discuss current limitations, futures research opportunities and potential broad impacts. △ Less

Submitted 13 October, 2021; originally announced October 2021.

Comments: Accepted at LWDA 21: Lernen, Wissen, Daten, Analysen September 2021, Munich, Germany