arXiv subjects

Christoph Clausen

Publications and source records attributed to Christoph Clausen.

11 recordsLinked to original sources

arXiv2026

Making the Discrete Continuous: Synthetic RAW Augmentations for Fine-Grained Evaluation of Person Detection Performance in Low Light

Real-world deployment of AI vision models is both fueled and limited by the data available for training and testing. Real datasets are sparse and uneven: long-tailed or unbalanced distributions hinder generalization, and the low number of samples in low density regions makes it hard to run evaluations. Synthetic data can fill these gaps, providing us with a way to sample the input space more continuously and improve data coverage for benchmarks. Focusing on the autonomous driving safety-critical case of pedestrian detection in the dark, we show how synthetic low-light samples can be used to better characterize the performance of a state-of-the-art object detection model as a function of the scene illumination. We use a synthetic RAW image augmentation technique to generate low-light samples that match the noise model of the camera sensor. Performance metrics on real and synthetic low-light data are similar, indicating that the AI model finds it hard to distinguish between them.

cs.CV↗

arXiv2023

Data Models for Dataset Drift Controls in Machine Learning With Optical Images

Camera images are ubiquitous in machine learning research. They also play a central role in the delivery of important services spanning medicine and environmental surveying. However, the application of machine learning models in these domains has been limited because of robustness concerns. A primary failure mode are performance drops due to differences between the training and deployment data. While there are methods to prospectively validate the robustness of machine learning models to such dataset drifts, existing approaches do not account for explicit models of the primary object of interest: the data. This limits our ability to study and understand the relationship between data generation and downstream machine learning model performance in a physically accurate manner. In this study, we demonstrate how to overcome this limitation by pairing traditional machine learning with physical optics to obtain explicit and differentiable data models. We demonstrate how such data models can be constructed for image data and used to control downstream machine learning model performance related to dataset drift. The findings are distilled into three applications. First, drift synthesis enables the controlled generation of physically faithful drift test cases to power model selection and targeted generalization. Second, the gradient connection between machine learning task model and data model allows advanced, precise tolerancing of task model sensitivity to changes in the data generation. These drift forensics can be used to precisely specify the acceptable data environments in which a task model may be run. Third, drift optimization opens up the possibility to create drifts that can help the task model learn better faster, effectively optimizing the data generating process itself. A guide to access the open code and datasets is available at https://github.com/aiaudit-org/raw2logit.

cs.LG↗

arXiv2020

Quantifying the effect of image compression on supervised learning applications in optical microscopy

The impressive growth of data throughput in optical microscopy has triggered a widespread use of supervised learning (SL) models running on compressed image datasets for efficient automated analysis. However, since lossy image compression risks to produce unpredictable artifacts, quantifying the effect of data compression on SL applications is of pivotal importance to assess their reliability, especially for clinical use. We propose an experimental method to evaluate the tolerability of image compression distortions in 2D and 3D cell segmentation SL tasks: predictions on compressed data are compared to the raw predictive uncertainty, which is numerically estimated from the raw noise statistics measured through sensor calibration. We show that predictions on object- and image-specific segmentation parameters can be altered by up to 15% and more than 10 standard deviations after 16-to-8 bits downsampling or JPEG compression. In contrast, a recently developed lossless compression algorithm provides a prediction spread which is statistically equivalent to that stemming from raw noise, while providing a compression ratio of up to 10:1. By setting a lower bound to the SL predictive uncertainty, our technique can be generalized to validate a variety of data analysis pipelines in SL-assisted fields.

eess.IV↗

arXiv2016

Fictitious magnetic field gradients in optical microtraps as an experimental tool for interrogating and manipulating cold atoms

Optical microtraps provide a strong spatial confinement for laser-cooled atoms. They can, e.g., be realized with strongly focused trapping light beams or the optical near fields of nano-scale waveguides and photonic nanostructures. Atoms in such traps often experience strongly spatially varying AC Stark shifts which are proportional to the magnetic quantum number of the respective energy level. These inhomogeneous fictitious magnetic fields can cause a displacement of the trapping potential that depends on the Zeeman state. Hitherto, this effect was mainly perceived as detrimental. However, it also provides a means to probe and to manipulate the motional state of the atoms in the trap by driving transitions between Zeeman states. Furthermore, by applying additional real or fictitious magnetic fields, the state-dependence of the trapping potential can be controlled. Here, using laser-cooled atoms that are confined in a nanofiber-based optical dipole trap, we employ this control in order to tune the microwave coupling of motional quantum states. We record corresponding microwave spectra which allow us to infer the trap parameters as well as the temperature of the atoms. Finally, we reduce the mean number of motional quanta in one spatial dimension to $\langle n\rangle=0.3 \pm 0.1$ by microwave sideband cooling. Our work shows that the inherent fictitious magnetic fields in optical microtraps expand the experimental toolbox for interrogating and manipulating cold atoms.

physics.atom-ph↗

arXiv2015

Directional spontaneous emission and lateral Casimir-Polder force on an atom close to a nanofiber

We study the spontaneous emission of an excited atom close to an optical nanofiber and the resulting scattering forces. For a suitably chosen orientation of the atomic dipole, the spontaneous emission pattern becomes asymmetric and a resonant Casimir--Polder force parallel to the fiber axis arises. For a simple model case, we show that the such a lateral force is due to the interaction of the circularly oscillating atomic dipole moment with its image inside the material. With the Casimir--Polder energy being constant in the lateral direction, the predicted lateral force does not derive from a potential in the usual way. Our results have implications for optical force measurements on a substrate as well as for laser cooling of atoms in nanophotonic traps.

quant-ph↗

arXiv2014

A source of polarization-entangled photon pairs interfacing quantum memories with telecom photons

We present a source of polarization-entangled photon pairs suitable for the implementation of long-distance quantum communication protocols using quantum memories. Photon pairs with wavelengths 883 nm and 1338 nm are produced by coherently pumping two periodically poled nonlinear waveguides embedded in the arms of a polarization interferometer. Subsequent spectral filtering reduces the bandwidth of the photons to 240 MHz. The bandwidth is well-matched to a quantum memory based on an Nd:YSO crystal, to which, in addition, the center frequency of the 883 nm photons is actively stabilized. A theoretical model that includes the effect of the filtering is presented and accurately fits the measured correlation functions of the generated photons. The model can also be used as a way to properly assess the properties of the source. The quality of the entanglement is revealed by a visibility of V = 96.1(9)% in a Bell-type experiment and through the violation of a Bell inequality.

quant-ph↗

arXiv2014

Quantum teleportation from a telecom-wavelength photon to a solid-state quantum memory

In quantum teleportation, the state of a single quantum system is disembodied into classical information and purely quantum correlations, to be later reconstructed onto a second system that has never directly interacted with the first one. This counterintuitive phenomenon is a cornerstone of quantum information science due to its essential role in several important tasks such as the long-distance transmission of quantum information using quantum repeaters. In this context, a challenge of paramount importance is the distribution of entanglement between remote nodes, and to use this entanglement as a resource for long-distance light-to-matter quantum teleportation. Here we demonstrate quantum teleportation of the polarization state of a telecom-wavelength photon onto the state of a solid-state quantum memory. Entanglement is established between a rare-earth-ion doped crystal storing a single photon that is polarization-entangled with a flying telecom-wavelength photon. The latter is jointly measured with another flying qubit carrying the polarization state to be teleported, which heralds the teleportation. The fidelity of the polarization state of the photon retrieved from the memory is shown to be greater than the maximum fidelity achievable without entanglement, even when the combined distances travelled by the two flying qubits is 25 km of standard optical fibre. This light-to-matter teleportation channel paves the way towards long-distance implementations of quantum networks with solid-state quantum memories.

quant-ph↗

arXiv2013

Analysis of a photon number resolving detector based on an ion Coulomb crystal inside an optical cavity

The ability to detect single photons with high efficiency is a crucial requirement for various quantum information applications. By combining the storage process of a quantum memory for photons with fluorescence-based quantum state measurement, it is in principle possible to achieve high efficiency photon counting in large ensembles of atoms. The large number of atoms can, however, pose significant problems in terms of noise stemming from imperfect initial state preparation and off-resonant fluorescence. We propose a concrete implementation of a photon number resolving detector based on an ion Coulomb crystal inside a moderately high-finesse optical cavity. The cavity enhancement leads to an effective optical depth of 15 for a finesse of 3000 with only about 1500 ions interacting with the light field. We show that these values allow for essentially noiseless detection with an efficiency larger than 90%. Moderate experimental parameters allow for repetition rates of about 5 kHz, limited by the time needed for fluorescence collection. Potential applications are discussed.

quant-ph↗

arXiv2012

Quantum storage of polarization qubits in birefringent and anisotropically absorbing materials

Storage of quantum information encoded into true single photons is an essential constituent of long-distance quantum communication based on quantum repeaters and of optical quantum information processing. The storage of photonic polarization qubits is, however, complicated by the fact that many materials are birefringent and have polarization-dependent absorption. Here we present and demonstrate a simple scheme that allows compensating for these polarization effects. The scheme is demonstrated using a solid-state quantum memory implemented with an ensemble of rare-earth ions doped into a biaxial yttrium orthosilicate ($Y_2SiO_5$) crystal. Heralded single photons generated from a filtered spontaneous parametric downconversion source are stored, and quantum state tomography of the retrieved polarization state reveals an average fidelity of $97.5 \pm 0.4%$, which is significantly higher than what is achievable with a measure-and-prepare strategy.

quant-ph↗

arXiv2011

Heralded quantum entanglement between two crystals

Quantum networks require the crucial ability to entangle quantum nodes. A prominent example is the quantum repeater which allows overcoming the distance barrier of direct transmission of single photons, provided remote quantum memories can be entangled in a heralded fashion. Here we report the observation of heralded entanglement between two ensembles of rare-earth-ions doped into separate crystals. A heralded single photon is sent through a 50/50 beamsplitter, creating a single-photon entangled state delocalized between two spatial modes. The quantum state of each mode is subsequently mapped onto a crystal, leading to an entangled state consisting of a single collective excitation delocalized between two crystals. This entanglement is revealed by mapping it back to optical modes and by estimating the concurrence of the retrieved light state. Our results highlight the potential of rare-earth-ions doped crystals for entangled quantum nodes and bring quantum networks based on solid-state resources one step closer.

quant-ph↗

arXiv2010

Quantum Storage of Photonic Entanglement in a Crystal

Entanglement is the fundamental characteristic of quantum physics. Large experimental efforts are devoted to harness entanglement between various physical systems. In particular, entanglement between light and material systems is interesting due to their prospective roles as "flying" and stationary qubits in future quantum information technologies, such as quantum repeaters and quantum networks. Here we report the first demonstration of entanglement between a photon at telecommunication wavelength and a single collective atomic excitation stored in a crystal. One photon from an energy-time entangled pair is mapped onto a crystal and then released into a well-defined spatial mode after a predetermined storage time. The other photon is at telecommunication wavelength and is sent directly through a 50 m fiber link to an analyzer. Successful transfer of entanglement to the crystal and back is proven by a violation of the Clauser-Horne-Shimony-Holt (CHSH) inequality by almost three standard deviations (S=2.64+/-0.23). These results represent an important step towards quantum communication technologies based on solid-state devices. In particular, our resources pave the way for building efficient multiplexed quantum repeaters for long-distance quantum networks.

quant-ph↗