Search

Objective: Enhancing the reliability of myoelectric controllers that decode motor intent is a pressing challenge in the field of bionic prosthetics. State-of-the-art research has mostly focused on Supervised Learning (SL) techniques to tackle this problem. However, obtaining high-quality labeled data that accurately represents muscle activity during daily usage remains difficult. We investigate the potential of Reinforcement Learning (RL) to further improve the decoding of human motion intent by incorporating usage-based data. Methods: The starting point of our method is a SL control policy, pretrained on a static recording of electromyographic (EMG) ground truth data. We then apply RL to fine-tune the pretrained classifier with dynamic EMG data obtained during interaction with a game environment developed for this work. We conducted real-time experiments to evaluate our approach and achieved significant improvements in human-in-the-loop performance. Results: The method effectively predicts simultaneous finger movements, leading to a two-fold increase in decoding accuracy during gameplay and a 39\% improvement in a separate motion test. Conclusion: By employing RL and incorporating usage-based EMG data during fine-tuning, our method achieves significant improvements in accuracy and robustness. Significance: These results showcase the potential of RL for enhancing the reliability of myoelectric controllers, of particular importance for advanced bionic limbs. See our project page for visual demonstrations: https://sites.google.com/view/bionic-limb-rl

Species distribution models (SDMs) aim to predict the distribution of species by relating occurrence data with environmental variables. Recent applications of deep learning to SDMs have enabled new avenues, specifically the inclusion of spatial data (environmental rasters, satellite images) as model predictors, allowing the model to consider the spatial context around each species' observations. However, the appropriate spatial extent of the images is not straightforward to determine and may affect the performance of the model, as scale is recognized as an important factor in SDMs. We develop a modular structure for SDMs that allows us to test the effect of scale in both single- and multi-scale settings. Furthermore, our model enables different scales to be considered for different modalities, using a late fusion approach. Results on the GeoLifeCLEF 2023 benchmark indicate that considering multimodal data and learning multi-scale representations leads to more accurate models., Comment: Published at the CV4Ecology workshop at ECCV 2024 (https://cv4e.netlify.app/papers/06.pdf)

Conjugation accelerates bacterial evolution by enabling bacteria to acquire genes horizontally from their neighbors. Because plasmid donors must physically connect with recipients to allow plasmid transfer, environmental fluid flows may increase conjugation rates by increasing cell-cell encounters through mixing driven by fluid shear, which creates relative movement between donors and recipients. However, existing experimental assays do not directly control cell-cell encounters, which hinders the establishment of a connection between the population-level conjugation rate and the microscale mechanisms that bring cells together. Here, we performed conjugation experiments between E. coli bacteria, varying the shear flow to control the rate of cell-cell encounters. We discovered that the conjugation rate increases with shear until it peaks at an optimal shear rate (approximately 100 1/s), reaching a value five-fold higher than the baseline set by diffusion-driven encounters. This optimum marks the transition from a regime in which shear promotes conjugation by increasing the rate of cell-cell encounters to a regime in which shear disrupts conjugation. Fluid flows are widely present in aquatic systems, gut, and soil, and our results indicate that fluid shear could induce hotspots of bacterial conjugation in the environment.

With the increasing demand for secure communication in optical space networks, it is essential to develop physical-layer scalable security solutions. In this context, we present the asymptotic security analysis of a keyless quantum private communication protocol that transmits classical information over quantum states. Different from the previous literature, our protocol sends dummy (decoy) states optimally obtained from the true information to deceive the eavesdropper. We analyze optical on-off keying (OOK) and binary phase shift keying (BPSK) for several detection scenarios. Our protocol significantly improves the protocol without decoy states whenever Bob is at a technological disadvantage with respect to Eve. Our protocol guarantees positive secrecy capacity when the eavesdropper gathers up to $90-99.9\%$ (depending on the detection scenario) of the photon energy that Bob detects, even when Eve is only limited by the laws of quantum mechanics. We apply our results to the design of an optical inter-satellite link (ISL) study case with pointing losses, and introduce a new design methodology whereby the link margin is guaranteed to be secure by our protocol. Hence, our design does not require knowing thr location of the eavesdropper and or channel state: the protocol aborts whenever the channel drops below the secured margin. Our protocol can be implemented with the state of the art space proof technology. Finally, we also show the potential secrecy advantage when using (not yet available) squeezed quantum states technology.

We show that the Morse boundary of a Morse local-to-global group is $\sigma$-compact. Moreover, we show that the converse holds for small cancellation groups. As an application, we show that the Morse boundary of a non-hyperbolic, Morse local-to-global group that has contraction does not admit a non-trivial stationary measure. In fact, we show that any stationary measure on a geodesic boundary of such a groups needs to assign measure zero to the Morse boundary. Unlike previous results, we do not need any assumptions on the stationary measures considered., Comment: 36 pages, 9 figures. Comments are welcome!

We show that the word problem for any 3-manifold group is solvable in time $O(n\log^3 n)$. Our main contribution is the proof that the word problem for admissible graphs of groups, in the sense of Croke and Kleiner, is solvable in $O(n\log n)$; this covers fundamental groups of non-geometric graph manifolds. Similar methods also give that the word problem for free products can be solved "almost as quickly" as the word problem in the factors., Comment: 24 pages, 1 figure

Since their inception, superconducting nanowire single-photon detectors have been enabling quantum optical applications and the rise of the photonic quantum industry. The evolution in the detector design and read-out strategies has led to the introduction of devices with a plurality of independent pixels, which have been able to operate with high system detection efficiency at high speed while also supporting photon number resolution capabilities. However, this comes at the cost of a complex readout that requires one coaxial cable for each pixel of the array. Here, we report a 28-pixel SNSPD with a dedicated parallel architecture that, while maintaining a simple readout with a single coaxial line, enables the detector to operate at high speed with low-performance degradation. The device shows a maximum single-photon efficiency (SDE) of 88% and is able to maintain its SDE above 50%, coupled with a timing jitter lower than 80~ps, up to a detection rate of 200 million counts per second (Mcps). The detector also provides state-of-the-art PNR performances with a 2-photon efficiency of 75% and a 3-photon efficiency of 60%.

We show that the Morse boundary exhibits interesting examples of both the existence and non-existence of Cannon-Thurston maps for normal subgroups, in contrast with the hyperbolic case., Comment: 7 pages, arguments streamlined and minor edits, accepted to Bulletin of the LMS

Given a geodesic metric space $X$, we construct a corresponding hyperbolic space, which we call the contraction space, that detects all strongly contracting directions in the following sense; a geodesic in $X$ is strongly contracting if and only if its parametrized image in the contraction space is a quasi-geodesic. If a finitely generated group $G$ acts geometrically on $X$, then all strongly-contracting elements act as WPD elements on the contraction space. If the space $X$ is CAT(0), or more generally Morse-dichotomous, that is if all Morse geodesics are strongly-contracting, then all generalized loxodromics act as WPD elements, implying that the action is what we call ``universally WPD''., Comment: 27 pages, 5 figures. Comments welcome!

In the face of significant biodiversity decline, species distribution models (SDMs) are essential for understanding the impact of climate change on species habitats by connecting environmental conditions to species occurrences. Traditionally limited by a scarcity of species observations, these models have significantly improved in performance through the integration of larger datasets provided by citizen science initiatives. However, they still suffer from the strong class imbalance between species within these datasets, often resulting in the penalization of rare species--those most critical for conservation efforts. To tackle this issue, this study assesses the effectiveness of training deep learning models using a balanced presence-only loss function on large citizen science-based datasets. We demonstrate that this imbalance-aware loss function outperforms traditional loss functions across various datasets and tasks, particularly in accurately modeling rare species with limited observations., Comment: Tackling Climate Change with Machine Learning at ICLR 2024

We study connected components of the Morse boundary and their stabilisers. We introduce the notion of point-convergence and show that if the set of non-singleton connected components of the Morse boundary of a finitely generated group $G$ is point-convergent, then every non-singleton connected component is the (relative) Morse boundary of its stabiliser. The above property only depends on the topology of the Morse boundary and hence is invariant under quasi-isometry. This shows that the topology of the Morse boundary not only carries algebraic information but can be used to detect certain subgroups which in some sense are invariant under quasi-isometry., Comment: 18 pages, 5 figures, comments welcome!

Species distribution modeling is a highly versatile tool for understanding the intricate relationship between environmental conditions and species occurrences. However, the available data often lacks information on confirmed species absence and is limited to opportunistically sampled, presence-only observations. To overcome this limitation, a common approach is to employ pseudo-absences, which are specific geographic locations designated as negative samples. While pseudo-absences are well-established for single-species distribution models, their application in the context of multi-species neural networks remains underexplored. Notably, the significant class imbalance between species presences and pseudo-absences is often left unaddressed. Moreover, the existence of different types of pseudo-absences (e.g., random and target-group background points) adds complexity to the selection process. Determining the optimal combination of pseudo-absences types is difficult and depends on the characteristics of the data, particularly considering that certain types of pseudo-absences can be used to mitigate geographic biases. In this paper, we demonstrate that these challenges can be effectively tackled by integrating pseudo-absences in the training of multi-species neural networks through modifications to the loss function. This adjustment involves assigning different weights to the distinct terms of the loss function, thereby addressing both the class imbalance and the choice of pseudo-absence types. Additionally, we propose a strategy to set these loss weights using spatial block cross-validation with presence-only data. We evaluate our approach using a benchmark dataset containing independent presence-absence data from six different regions and report improved results when compared to competing approaches.

Distributing entangled states over potentially long distances provides a key resource for many protocols in quantum communication and quantum cryptography. Ideally, this should be implemented in a heralded manner. By starting with four single-photon states, we cascade two single-photon path-entangled states, coded in orthogonal polarizations, to distribute and herald polarization entanglement in a single quantum repeater link architecture. By tuning the input states to minimize (local) losses, the theoretically achievable fidelity to the target state without postselection approaches 1, while sacrificing heralding rates. We achieve a fidelity to the target state of over 95% after postselection, providing a benchmark for the experimental control. We show that the fidelity of the heralded state without postselection scales predictably and also identify various practical challenges and error sources specific to this architecture, and model their effects on the generated state. While our experiment uses probabilistic photon-pair sources based on spontaneous parametric down-conversion, many of these problems are also relevant for variants employing deterministic photon sources.

Learning representations of geographical space is vital for any machine learning model that integrates geolocated data, spanning application domains such as remote sensing, ecology, or epidemiology. Recent work embeds coordinates using sine and cosine projections based on Double Fourier Sphere (DFS) features. These embeddings assume a rectangular data domain even on global data, which can lead to artifacts, especially at the poles. At the same time, little attention has been paid to the exact design of the neural network architectures with which these functional embeddings are combined. This work proposes a novel location encoder for globally distributed geographic data that combines spherical harmonic basis functions, natively defined on spherical surfaces, with sinusoidal representation networks (SirenNets) that can be interpreted as learned Double Fourier Sphere embedding. We systematically evaluate positional embeddings and neural network architectures across various benchmarks and synthetic evaluation datasets. In contrast to previous approaches that require the combination of both positional encoding and neural networks to learn meaningful representations, we show that both spherical harmonics and sinusoidal representation networks are competitive on their own but set state-of-the-art performances across tasks when combined. The model code and experiments are available at https://github.com/marccoru/locationencoder., Comment: Camera-ready version

We introduce the notion of mixed subtree quasi-isometries, which are self quasi-isometries of regular trees built in a specific inductive way. We then show that any self quasi-isometry of a regular tree is at bounded distance from a mixed-subtree quasi-isometry. Since the free group is quasi-isometric to a regular tree, this provides a way to describe all self quasi-isometries of the free group. In doing this, we also give a way of constructing quasi-isometries of the free group., Comment: Modified the proof structure in Section 3 for better readability. Accepted in Algebraic & Geometric Topology

We provide examples of classical small-cancellation groups which have non-sigma-compact Morse boundary. These are first known examples of groups with non-sigma-compact Morse boundary. Some small-cancellation groups do have sigma-compact Morse boundary, so this property distinguishes quasi-isometry types of small-cancellation groups. In fact, we give a complete description of when Morse boundaries of $C'(1/6)$--groups have sigma-compact Morse boundary. We also provide examples of $C'(1/6)$--groups where all Morse rays are strongly contracting., Comment: Comments welcome!

Random walks cannot, in general, be pushed forward by quasi-isometries. Tame Markov chains were introduced as a `quasi-isometry invariant' are a generalization of random walks. In this paper, we construct several examples of tame Markov chains on the free group exhibiting `exotic' behaviour; one, where the drift is not well defined and one where the drift is well defined but the Central Limit Theorem does not hold. We show that this is not a failure of the notion of tame Markov chain, but rather that any quasi-isometry invariant theory that generalizes random walks will include examples without well-defined drift., Comment: 15 pages, 2 figures. Corrected some typos and added minor clarifications. To appear in Journal of Theoretical Probability

Superconducting-nanowire single-photon detectors (SNSPDs) have enabled the realization of several quantum optics technologies thanks to their high detection efficiency, low dark-counts, and fast recovery time. However, the widespread use of technologies such as linear optical quantum computing (LOQC), quasi-deterministic single photon sources and quantum repeaters requires faster detectors that can distinguish between different photon number states. Here, we report the fabrication of an SNSPD array composed of 14 independent pixels, achieving a system detection efficiency (SDE) of 90% in the telecom band. By reading each pixel of the array independently we show that the detector can detect telecom photons at 1.5 GHz with 45% absolute SDE. We exploit the dynamic PNR of the array to demonstrate accurate state reconstruction for different photon-number statistics for a wide range of light inputs, including operation with long-duration light pulses, as commonly obtained with some cavity-based sources. We show 2-photon and 3-photon fidelities of 74% and 57% respectively, which represent state-of-the-art results for fiber-coupled SNSPDs.

We consider cusped hyperbolic $n-$manifolds, and compute \v{C}ech cohomology groups of the Morse boundaries of their fundamental groups. In particular, we show that the reduced \v{C}ech cohomology with real coefficients vanishes in dimension at most $n-3$ and does not vanish in dimension $n-2$. A similar result holds for relatively hyperbolic groups with virtually nilpotent peripherals and Bowditch boundary homeomorphic to a sphere; these include all non-uniform lattices in rank$-1$ simple Lie groups., Comment: 41 pages, 8 figures; minor edits, accepted in Advances in Mathematics

Semantic segmentation has made significant progress in recent years thanks to deep neural networks, but the common objective of generating a single segmentation output that accurately matches the image's content may not be suitable for safety-critical domains such as medical diagnostics and autonomous driving. Instead, multiple possible correct segmentation maps may be required to reflect the true distribution of annotation maps. In this context, stochastic semantic segmentation methods must learn to predict conditional distributions of labels given the image, but this is challenging due to the typically multimodal distributions, high-dimensional output spaces, and limited annotation data. To address these challenges, we propose a conditional categorical diffusion model (CCDM) for semantic segmentation based on Denoising Diffusion Probabilistic Models. Our model is conditioned to the input image, enabling it to generate multiple segmentation label maps that account for the aleatoric uncertainty arising from divergent ground truth annotations. Our experimental results show that CCDM achieves state-of-the-art performance on LIDC, a stochastic semantic segmentation dataset, and outperforms established baselines on the classical segmentation dataset Cityscapes., Comment: Accepted at ICCV 2023. Code available at https://github.com/LarsDoorenbos/ccdm-stochastic-segmentation

The prediction of solar flares is of practical and scientific interest; however, many machine learning methods used for this prediction task do not provide the physical explanations behind a model's performance. We made use of two recently developed explainable artificial intelligence techniques called gradient-weighted class activation mapping (Grad-CAM) and expected gradients (EG) to reveal the decision-making process behind a high-performance neural network that has been trained to distinguish between MgII spectra derived from flaring and nonflaring active regions, a fact that can be applied to the task of short timescale flare forecasting. The two techniques generate visual explanations (heatmaps) that can be projected back onto the spectra, allowing for the identification of features that are strongly associated with precursory flare activity. We automated the search for explainable interpretations on the level of individual wavelengths, and provide multiple examples of flare prediction using IRIS spectral data, finding that prediction scores in general increase before flare onset. Large IRIS rasters that cover a significant portion of the active region and coincide with small preflare brightenings both in IRIS and SDO/AIA images tend to lead to better forecasts. The models reveal that MgII triplet emission, flows, as well as broad and highly asymmetric spectra are all important for the task of flare prediction. Additionally, we find that intensity is only weakly correlated to a spectrum's prediction score, meaning that low intensity spectra can still be of great importance for the flare prediction task, and that $78$% of the time, the position of the model's maximum attention along the slit during the preflare phase is predictive of the location of the flare's maximum UV emission

We classify the homeomorphism types of the Morse boundaries of all 3-manifold groups into 9 different possible homeomorphism types, and show how the Morse boundary depends on the geometric decomposition., Comment: to be published in Transactions of the American Mathematical Society

Ever since the creation of the first artificial intelligence (AI) machinery built on machine learning (ML), public society has entertained the idea that eventually computers could become sentient and develop a consciousness of their own. As these models now get increasingly better and convincingly more anthropomorphic, even some engineers have started to believe that AI might become conscious, which would result in serious social consequences. The present paper argues against the plausibility of sentient AI primarily based on the theory of neurogenetic structuralism, which claims that the physiology of biological neurons and their structural organization into complex brains are necessary prerequisites for true consciousness to emerge.

Quantum key distribution (QKD) is nowadays a well established method for generating secret keys at a distance in an information-theoretic secure way, as the secrecy of QKD relies on the laws of quantum physics and not computational complexity. In order to industrialize QKD, low-cost, mass-manufactured and practical QKD setups are required. Hence, photonic and electronic integration of the sender's and receiver's respective components is currently in the spotlight. Here we present a high-speed (2.5 GHz) integrated QKD setup featuring a transmitter chip in silicon photonics allowing for high-speed modulation and accurate state preparation, as well as a polarization-independent low-loss receiver chip in aluminum borosilicate glass fabricated by the femtosecond laser micromachining technique. Our system achieves raw bit error rates, quantum bit error rates and secret key rates equivalent to a much more complex state-of-the-art setup based on discrete components.

This work demonstrates the capabilities of an entangled photon-pair source at telecom wavelengths, based on a photonic integrated Si$_3$N$_4$ microresonator with monolithically integrated piezoelectric frequency tuning. Previously, frequency tuning of photon-pairs generated by microresonators has only been demonstrated using thermal control, however these have limited actuation bandwidth, and are not compatible with cryogenic environments. Here, the frequency-tunable photon-pair generation capabilities of a Si$_3$N$_4$ microresonator with a monolithically integrated aluminium nitride layer are shown. Fast-frequency locking of the microresonator to an external laser is demonstrated, with a resulting locking bandwidth orders of magnitude larger than reported previously using thermal locking. These abilities will have direct application in future schemes which interface such sources with quantum memories based on e.g. trapped-ion or rare-earth ion schemes.

Quantum Key Distribution has made continuous progress over the last 20 years and is now commercially available. However, the secret key rates (SKR) are still limited to a few Mbps. Here, we present a custom multipixel superconducting nanowire single-photon detectors and fast acquisition and real-time key distillation electronics, removing two roadblocks and allowing an increase of the SKR of more than an order of magnitude. In combination with a simple 2.5 GHz clocked time-bin quantum key distribution system, we can generate secret keys at a rate of 64 Mbps over a distance of 10.0 km and at a rate of 3.0 Mbps over a distance of 102.4 km with real-time key distillation., Comment: 5 pages, 5 figures, submitted to Nature Photonics

Heralded single-photon sources (HSPS) intrinsically suffer from multiphoton emission, leading to a trade-off between the source's quality and the heralding rate. A solution to this problem is to use photon-number-resolving (PNR) detectors to filter out the heralding events where more than one photon pair is created. Here, we demonstrate the use of a high-efficiency PNR superconducting nanowire single-photon detector (SNSPD) as a heralding detector for a HSPS. By filtering out higher-order heralding detections, we can reduce the $g^{(2)}(0)$ of the heralded single photon by $(26.6 \pm 0.2)\,\%$, or alternatively, for a fixed pump power, increasing the heralding rate by a factor of $1.363 \pm 0.004$ for a fixed $g^{(2)}(0)$. Additionally, we use the detector to directly measure the photon-number distribution of a thermal mode and calculate the unheralded $g^{(2)}(0)$. We show the possibility to perform $g^{(2)}(0)$ measurements with only one PNR detector, with the results in agreement with those obtained by more common-place techniques which use multiple threshold detectors. Our work shows that efficient PNR SNSPDs can significantly improve the performance of HSPSs and can precisely characterize them, making these detectors a useful tool for a wide range of optical quantum information protocols.

A recently published patent (https://www.ipo.gov.uk/p-ipsum/Case/PublicationNumber/GB2590064) has claimed the development of a novel quantum key distribution protocol purporting to achieve long-range quantum security without trusted nodes and without use of quantum repeaters. Here we present a straightforward analysis of this claim, and reach the conclusion that it is largely unfounded., Comment: 9 pages, 2 figures and 1 table

Taking advantage of the many recent advances in deep learning, text-to-image generative models currently have the merit of attracting the general public attention. Two of these models, DALL-E 2 and Imagen, have demonstrated that highly photorealistic images could be generated from a simple textual description of an image. Based on a novel approach for image generation called diffusion models, text-to-image models enable the production of many different types of high resolution images, where human imagination is the only limit. However, these models require exceptionally large amounts of computational resources to train, as well as handling huge datasets collected from the internet. In addition, neither the codebase nor the models have been released. It consequently prevents the AI community from experimenting with these cutting-edge models, making the reproduction of their results complicated, if not impossible. In this thesis, we aim to contribute by firstly reviewing the different approaches and techniques used by these models, and then by proposing our own implementation of a text-to-image model. Highly based on DALL-E 2, we introduce several slight modifications to tackle the high computational cost induced. We thus have the opportunity to experiment in order to understand what these models are capable of, especially in a low resource regime. In particular, we provide additional and analyses deeper than the ones performed by the authors of DALL-E 2, including ablation studies. Besides, diffusion models use so-called guidance methods to help the generating process. We introduce a new guidance method which can be used in conjunction with other guidance methods to improve image quality. Finally, the images generated by our model are of reasonably good quality, without having to sustain the significant training costs of state-of-the-art text-to-image models., Comment: Master's Thesis

Photon-number resolving (PNR) single-photon detectors are an enabling technology in many areas such as photonic quantum computing, non-classical light source characterisation and quantum imaging. Here, we demonstrate high-efficiency PNR detectors using a parallel superconducting nanowire single-photon detector (P-SNSPD) architecture that does not suffer from crosstalk between the pixels and that is free of latching. The behavior of the detector is modelled and used to predict the possible outcomes given a certain number of incoming photons. We apply our model to a 4-pixel P-SNSPD with a system detection efficiency of 92.5%. We also demonstrate how this detector allows reconstructing the photon-number statistics of a coherent source of light, which paves the way towards the characterisation of the photon statistics of other types of light source using a single detector., Comment: 8 pages, 7 figures

We experimentally study entangled two-photon absorption in Rhodamine 6G as a function of the spatial properties of a high flux of broadband entangled photon pairs. We first demonstrate a key signature dependence of the entangled two-photon absorption rate on the type of entangled pair flux attenuation: linear, when the laser pump power is attenuated, and quadratic, when the pair flux itself experiences linear loss. We then perform a fluorescence-based Z-scan measurement to study the influence of beam waist size on the entangled two-photon absorption process and compare this to classical single- and two-photon absorption processes. We demonstrate that the entangled two-photon absorption shares a beam waist dependence similar to that of classical two-photon absorption. This result presents an additional argument for the wide range of contrasting values of quoted entangled two-photon absorption cross-sections of dyes in literature.