Search

Recently, deep learning has experienced rapid expansion, contributing significantly to the progress of supervised learning methodologies. However, acquiring labeled data in real-world settings can be costly, labor-intensive, and sometimes scarce. This challenge inhibits the extensive use of neural networks for practical tasks due to the impractical nature of labeling vast datasets for every individual application. To tackle this, semi-supervised learning (SSL) offers a promising solution by using both labeled and unlabeled data to train object detectors, potentially enhancing detection efficacy and reducing annotation costs. Nevertheless, SSL faces several challenges, including pseudo-target inconsistencies, disharmony between classification and regression tasks, and efficient use of abundant unlabeled data, especially on edge devices, such as roadside cameras. Thus, we developed a teacher-student-based SSL framework, Co-Learning, which employs mutual learning and annotation-alignment strategies to adeptly navigate these complexities and achieves comparable performance as fully-supervised solutions using 10\% labeled data., Comment: Accepted at EAmSI24: Edge AI meets swarm intelligence

This paper presents a preliminary study of an efficient object tracking approach, comparing the performance of two different 3D point cloud sensory sources: LiDAR and stereo cameras, which have significant price differences. In this preliminary work, we focus on single object tracking. We first developed a fast heuristic object detector that utilizes prior information about the environment and target. The resulting target points are subsequently fed into an extended object tracking framework, where the target shape is parameterized using a star-convex hypersurface model. Experimental results show that our object tracking method using a stereo camera achieves performance similar to that of a LiDAR sensor, with a cost difference of more than tenfold.

In this paper, we present an experimental comparison of various graph-based approximate nearest neighbor (ANN) search algorithms deployed on edge devices for real-time nearest neighbor search applications, such as smart city infrastructure and autonomous vehicles. To the best of our knowledge, this specific comparative analysis has not been previously conducted. While existing research has explored graph-based ANN algorithms, it has often been limited to single-threaded implementations on standard commodity hardware. Our study leverages the full computational and storage capabilities of edge devices, incorporating additional metrics such as insertion and deletion latency of new vectors and power consumption. This comprehensive evaluation aims to provide valuable insights into the performance and suitability of these algorithms for edge-based real-time tracking systems enhanced by nearest-neighbor search algorithms.

Search and rescue operations require mobile robots to navigate unstructured indoor and outdoor environments. In particular, actively stabilized multirotor drones need precise movement data to balance and avoid obstacles. Combining radial velocities from on-chip radar with MEMS inertial sensing has proven to provide robust, lightweight, and consistent state estimation, even in visually or geometrically degraded environments. Statistical tests robustify these estimators against radar outliers. However, available work with binary outlier filters lacks adaptability to various hardware setups and environments. Other work has predominantly been tested in handheld static environments or automotive contexts. This work introduces a robust baro-radar-inertial odometry (BRIO) m-estimator for quadcopter flights in typical GNSS-denied scenarios. Extensive real-world closed-loop flights in cities and forests demonstrate robustness to moving objects and ghost targets, maintaining a consistent performance with 0.5 % to 3.2 % drift per distance traveled. Benchmarks on public datasets validate the system's generalizability. The code, dataset, and video are available at https://github.com/ethz-asl/rio., Comment: Accepted for publication at IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI) 2024

While significant progress has been made on the hardware side of quantum computing, support for high-level quantum programming abstractions remains underdeveloped compared to classical programming languages. In this article, we introduce Qrisp, a framework designed to bridge several gaps between high-level programming paradigms in state-of-the-art software engineering and the physical reality of today's quantum hardware. The framework aims to provide a systematic approach to quantum algorithm development such that they can be effortlessly implemented, maintained and improved. We propose a number of programming abstractions that are inspired by classical paradigms, yet consistently focus on the particular needs of a quantum developer. Unlike many other high-level language approaches, Qrisp's standout feature is its ability to compile programs to the circuit level, making them executable on most existing physical backends. The introduced abstractions enable the Qrisp compiler to leverage algorithm structure for increased compilation efficiency. Finally, we present a set of code examples, including an implementation of Shor's factoring algorithm. For the latter, the resulting circuit shows significantly reduced quantum resource requirements, strongly supporting the claim that systematic quantum algorithm development can give quantitative benefits.

Unsupervised Domain Adaptation (UDA) has shown significant advancements in object detection under well-lit conditions; however, its performance degrades notably in low-visibility scenarios, especially at night, posing challenges not only for its adaptability in low signal-to-noise ratio (SNR) conditions but also for the reliability and efficiency of automated vehicles. To address this problem, we propose a \textbf{Co}operative \textbf{S}tudents (\textbf{CoS}) framework that innovatively employs global-local transformations (GLT) and a proxy-based target consistency (PTC) mechanism to capture the spatial consistency in day- and night-time scenarios effectively, and thus bridge the significant domain shift across contexts. Building upon this, we further devise an adaptive IoU-informed thresholding (AIT) module to gradually avoid overlooking potential true positives and enrich the latent information in the target domain. Comprehensive experiments show that CoS essentially enhanced UDA performance in low-visibility conditions and surpasses current state-of-the-art techniques, achieving an increase in mAP of 3.0\%, 1.9\%, and 2.5\% on BDD100K, SHIFT, and ACDC datasets, respectively. Code is available at https://github.com/jichengyuan/Cooperitive_Students., Comment: Code is available at https://github.com/jichengyuan/Cooperitive_Students

The quantum backtracking algorithm proposed by Ashley Montanaro raised considerable interest, as it provides a quantum speed-up for a large class of classical optimization algorithms. It does not suffer from Barren-Plateaus and transfers well into the fault-tolerant era, as it requires only a limited number of arbitrary angle gates. Despite its potential, the algorithm has seen limited implementation efforts, presumably due to its abstract formulation. In this work, we provide a detailed instruction on implementing the quantum step operator for arbitrary backtracking instances. For a single controlled diffuser of a binary backtracking tree with depth n, our implementation requires only $6n+14$ CX gates. We detail the process of constructing accept and reject oracles for Sudoku problems using our interface to quantum backtracking. The presented code is written using Qrisp, a high-level quantum programming language, making it executable on most current physical backends and simulators. Subsequently, we perform several simulator based experiments and demonstrate solving 4x4 Sudoku instances with up to 9 empty fields. This is, to the best of our knowledge, the first instance of a compilable implementation of this generality, marking a significant and exciting step forward in quantum software engineering.

The availability of vast amounts of visual data with heterogeneous features is a key factor for developing, testing, and benchmarking of new computer vision (CV) algorithms and architectures. Most visual datasets are created and curated for specific tasks or with limited image data distribution for very specific situations, and there is no unified approach to manage and access them across diverse sources, tasks, and taxonomies. This not only creates unnecessary overheads when building robust visual recognition systems, but also introduces biases into learning systems and limits the capabilities of data-centric AI. To address these problems, we propose the Vision Knowledge Graph (VisionKG), a novel resource that interlinks, organizes and manages visual datasets via knowledge graphs and Semantic Web technologies. It can serve as a unified framework facilitating simple access and querying of state-of-the-art visual datasets, regardless of their heterogeneous formats and taxonomies. One of the key differences between our approach and existing methods is that ours is knowledge-based rather than metadatabased. It enhances the enrichment of the semantics at both image and instance levels and offers various data retrieval and exploratory services via SPARQL. VisionKG currently contains 519 million RDF triples that describe approximately 40 million entities, and are accessible at https://vision.semkg.org and through APIs. With the integration of 30 datasets and four popular CV tasks, we demonstrate its usefulness across various scenarios when working with CV pipelines., Comment: Accepted at ESWC 2024

Uncomputation is an essential part of reversible computing and plays a vital role in quantum computing. Using this technique, memory resources can be safely deallocated without performing a nonreversible deletion process. For the case of quantum computing, several algorithms depend on this as they require disentangled states in the course of their execution. Thus, uncomputation is not only about resource management, but is also required from an algorithmic point of view. However, synthesizing uncomputation circuits is tedious and can be automated. In this paper, we describe the interface for automated generation of uncomputation circuits in our Qrisp framework. Our algorithm for synthesizing uncomputation circuits in Qrisp is based on an improved version of "Unqomp", a solution presented by Paradis et. al. Our paper also presents some improvements to the original algorithm, in order to make it suitable for the needs of a high-level programming framework. Qrisp itself is a fully compilable, high-level programming language/framework for gate-based quantum computers, which abstracts from many of the underlying hardware details. Qrisp's goal is to support a high-level programming paradigm as known from classical software development.

Quantum information processing architectures typically only allow for nearest-neighbour entanglement creation. In many cases, this prevents the direct generation of GHZ states, which are commonly used for many communication and computation tasks. Here, we show how to obtain GHZ states between nodes in a network that are connected in a straight line, naturally allowing them to initially share linear cluster states. We prove a strict upper bound of $\lfloor (n+3)/2 \rfloor$ on the size of the set of nodes sharing a GHZ state that can be obtained from a linear cluster state of $n$ qubits, using local Clifford unitaries, local Pauli measurements, and classical communication. Furthermore, we completely characterize all selections of nodes below this threshold that can share a GHZ state obtained within this setting. Finally, we demonstrate these transformations on the IBMQ Montreal quantum device for linear cluster states of up to $n=19$ qubits., Comment: 6 pages, 4 figures, Jupyter notebook available at https://github.com/hahnfrederik/Extracting-maximal-entanglement-from-linear-cluster-states

We construct a family of compact free boundary minimal annuli immersed in the unit ball $\mathbb{B}^3$ of $\mathbb{R}^3$, the first such examples other than the critical catenoid. This solves a problem formulated by Nitsche in 1985. These annuli are symmetric with respect to two orthogonal planes and a finite group of rotations around an axis, and are foliated by spherical curvature lines. We show that the only free boundary minimal annulus embedded in $\mathbb{B}^3$ foliated by spherical curvature lines is the critical catenoid; in particular, the minimal annuli that we construct are not embedded. On the other hand, we also construct families of non-rotational compact embedded capillary minimal annuli in $\mathbb{B}^3$. Their existence solves in the negative a problem proposed by Wente in 1995., Comment: 39 pages, 11 figures. We added a uniqueness section, we explained in more detail several arguments, and introduced new figures. The existence theorems of version 1 remain unchanged

We present CQELS 2.0, the second version of Continuous Query Evaluation over Linked Streams. CQELS 2.0 is a platform-agnostic federated execution framework towards semantic stream fusion. In this version, we introduce a novel neural-symbolic stream reasoning component that enables specifying deep neural network (DNN) based data fusion pipelines via logic rules with learnable probabilistic degrees as weights. As a platform-agnostic framework, CQELS 2.0 can be implemented for devices with different hardware architectures (from embedded devices to cloud infrastructures). Moreover, this version also includes an adaptive federator that allows CQELS instances on different nodes in a network to coordinate their resources to distribute processing pipelines by delegating partial workloads to their peers via subscribing continuous queries

Stream processing and reasoning is getting considerable attention in various application domains such as IoT, Industry IoT and Smart Cities. In parallel, reasoning and knowledge-based features have attracted research into many areas of robotics, such as robotic mapping, perception and interaction. To this end, the Semantic Stream Reasoning (SSR) framework can unify the representations of symbolic/semantic streams with deep neural networks, to integrate high-dimensional data streams, such as video streams and LiDAR point clouds, with traditional graph or relational stream data. As such, this positioning and system paper will outline our approach to build a platform to facilitate semantic stream reasoning capabilities on a robotic operating system called SemRob.

One of the major promises of quantum computing is the realization of SIMD (single instruction - multiple data) operations using the phenomenon of superposition. Since the dimension of the state space grows exponentially with the number of qubits, we can easily reach situations where we pay less than a single quantum gate per data point for data-processing instructions which would be rather expensive in classical computing. Formulating such instructions in terms of quantum gates, however, still remains a challenging task. Laying out the foundational functions for more advanced data-processing is therefore a subject of paramount importance for advancing the realm of quantum computing. In this paper, we introduce the formalism of encoding so called-semi-boolean polynomials. As it turns out, arithmetic $\mathbb{Z}/2^n\mathbb{Z}$ ring operations can be formulated as semi-boolean polynomial evaluations, which allows convenient generation of unsigned integer arithmetic quantum circuits. For arithmetic evaluations, the resulting algorithm has been known as Fourier-arithmetic. We extend this type of algorithm with additional features, such as ancilla-free in-place multiplication and integer coefficient polynomial evaluation. Furthermore, we introduce a tailor-made method for encoding signed integers succeeded by an encoding for arbitrary floating-point numbers. This representation of floating-point numbers and their processing can be applied to any quantum algorithm that performs unsigned modular integer arithmetic. We discuss some further performance enhancements of the semi boolean polynomial encoder and finally supply a complexity estimation. The application of our methods to a 32-bit unsigned integer multiplication demonstrated a 90\% circuit depth reduction compared to carry-ripple approaches.

The steadily growing research interest in quantum computing - together with the accompanying technological advances in the realization of quantum hardware - fuels the development of meaningful real-world applications, as well as implementations for well-known quantum algorithms. One of the most prominent examples till today is Grover's algorithm, which can be used for efficient search in unstructured databases. Quantum oracles that are frequently masked as black boxes play an important role in Grover's algorithm. Hence, the automatic generation of oracles is of paramount importance. Moreover, the automatic generation of the corresponding circuits for a Grover quantum oracle is deeply linked to the synthesis of reversible quantum logic, which - despite numerous advances in the field - still remains a challenge till today in terms of synthesizing efficient and scalable circuits for complex boolean functions. In this paper, we present a flexible method for automatically encoding unstructured databases into oracles, which can then be efficiently searched with Grover's algorithm. Furthermore, we develop a tailor-made method for quantum logic synthesis, which vastly improves circuit complexity over other current approaches. Finally, we present another logic synthesis method that considers the requirements of scaling onto real world backends. We compare our method with other approaches through evaluating the oracle generation for random databases and analyzing the resulting circuit complexities using various metrics.

We construct complete, embedded minimal annuli asymptotic to vertical planes in the Riemannian 3-manifold PSL. The boundary of these annuli consists of 4 vertical lines at infinity. They are constructed by taking the limit of a sequence of compact minimal annuli. The compactness is obtained from an estimate of curvature which uses foliations by minimal surfaces. This estimate is independent of the index of the surface. We also prove the existence of a one-periodic family of Riemann's type examples. The difficulty of the construction comes from the lack of symmetry of the ambient space PSL.

We prove that a properly embedded annular end of a surface in $\mathbb H^2\times\mathbb R$ with constant mean curvature $0

Mathematical optimization is one of the cornerstones of modern engineering research and practice. Yet, throughout all application domains, mathematical optimization is, for the most part, considered to be a numerical discipline. Optimization problems are formulated to be solved numerically with specific algorithms running on microprocessors. An emerging alternative is to view optimization algorithms as dynamical systems. Besides being insightful in itself, this perspective liberates optimization methods from specific numerical and algorithmic aspects and opens up new possibilities to endow complex real-world systems with sophisticated self-optimizing behavior. Towards this goal, it is necessary to understand how numerical optimization algorithms can be converted into feedback controllers to enable robust "closed-loop optimization". In this article, we focus on recent control designs under the name of "feedback-based optimization" which implement optimization algorithms directly in closed loop with physical systems. In addition to a brief overview of selected continuous-time dynamical systems for optimization, our particular emphasis in this survey lies on closed-loop stability as well as the robust enforcement of physical and operational constraints in closed-loop implementations. To bypass accessing partial model information of physical systems, we further elaborate on fully data-driven and model-free operations. We highlight an emerging application in autonomous reserve dispatch in power systems, where the theory has transitioned to practice by now. We also provide short expository reviews of pioneering applications in communication networks and electricity grids, as well as related research streams, including extremum seeking and pertinent methods from model predictive and process control, to facilitate high-level comparisons with the main topic of this survey.

In this paper, we study the stability and convergence of continuous-time Lagrangian saddle flows to solutions of a convex constrained optimization problem. Convergence of these flows is well-known when the underlying saddle function is either strictly convex in the primal or strictly concave in the dual variables. In this paper, we show convergence under non-strict convexity when a simple, unilateral augmentation term is added. For this purpose, we establish a novel, non-trivial characterization of the limit set of saddle-flow trajectories that allows us to preclude limit cycles. With our presentation we try to unify several existing problem formulations as a projected dynamical system that allows projection of both the primal and dual variables, thus complementing results available in the recent literature.

Inspired by classical sensitivity results for nonlinear optimization, we derive and discuss new quantitative bounds to characterize the solution map and dual variables of a parametrized nonlinear program. In particular, we derive explicit expressions for the local and global Lipschitz constants of the solution map of non-convex or convex optimization problems, respectively. Our results are geared towards the study of time-varying optimization problems which are commonplace in various applications of online optimization, including power systems, robotics, signal processing and more. In this context, our results can be used to bound the rate of change of the optimizer. To illustrate the use of our sensitivity bounds we generalize existing arguments to quantify the tracking performance of continuous-time, monotone running algorithms. Further, we introduce a new continuous-time running algorithm for time-varying constrained optimization which we model as a so-called perturbed sweeping process. For this discontinuous scheme, we establish an explicit bound on the asymptotic solution tracking for a class of convex problems.

The publication and (re)utilization of Open Data is still facing multiple barriers on technical, organizational and legal levels. This includes limitations in interfaces, search capabilities, provision of quality information and the lack of definite standards and implementation guidelines. Many Semantic Web specifications and technologies are specifically designed to address the publication of data on the web. In addition, many official publication bodies encourage and foster the development of Open Data standards based on Semantic Web principles. However, no existing solution for managing Open Data takes full advantage of these possibilities and benfits. In this paper, we present our solution "Piveau", a fully-fledged Open Data management solution, based on Semantic Web technologies. It harnesses a variety of standards, like RDF, DCAT, DQV, and SKOS, to overcome the barriers in Open Data publication. The solution puts a strong focus on assuring data quality and scalability. We give a detailed description of the underlying, highly scalable, service-oriented architecture, how we integrated the aforementioned standards, and used a triplestore as our primary database. We have evaluated our work in a comprehensive feature comparison to established solutions and through a practical application in a production environment, the European Data Portal. Our solution is available as Open Source., Comment: 16 pages, 2 figures, preprint of an in-use paper at Extended Semantic Web Conference (ESWC 2020)

In this paper, we present a novel control scheme for feedback optimization. That is, we propose a discrete-time controller that can steer the steady state of a physical plant to the solution of a constrained optimization problem without numerically solving the problem. Our controller can be interpreted as a discretization of a continuous-time projected gradient flow. Compared to other schemes used for feedback optimization, such as saddle-point flows or inexact penalty methods, our algorithm combines several desirable properties: It asymptotically enforces constraints on the plant steady-state outputs, and temporary constraint violations can be easily quantified. Our algorithm requires only reduced model information in the form of steady-state input-output sensitivities of the plant. Further, as we prove in this paper, global convergence is guaranteed even for non-convex problems. Finally, our algorithm is straightforward to tune, since the step-size is the only tuning parameter., Comment: 6 pages, 3 figures

This paper concerns a new class of discontinuous dynamical systems for constrained optimization. These dynamics are particularly suited to solve nonlinear, non-convex problems in closed-loop with a physical system. Such approaches using feedback controllers that emulate optimization algorithms have recently been proposed for the autonomous optimization of power systems and other infrastructures. In this paper, we consider feedback gradient flows that exploit physical input saturation with the help of anti-windup control to enforce constraints. We prove semi-global convergence of "projected" trajectories to first-order optimal points, i.e., of the trajectories obtained from a pointwise projection onto the feasible set. In the process, we establish properties of the directional derivative of the projection map for non-convex, prox-regular sets.

In this paper we study how high-gain anti-windup schemes can be used to implement projected dynamical systems in control loops that are subject to saturation on a (possibly unknown) set of admissible inputs. This insight is especially useful for the design of autonomous optimization schemes that realize a closed-loop behavior which approximates a particular optimization algorithm (e.g., projected gradient or Newton descent) while requiring only limited model information. In our analysis we show that a saturated integral controller, augmented with an anti-windup scheme, gives rise to a perturbed projected dynamical system. This insight allows us to show uniform convergence and robust practical stability as the anti-windup gain goes to infinity. Moreover, for a special case encountered in autonomous optimization we show robust convergence, i.e., convergence to an optimal steady-state for finite gains. Apart from being particularly suited for online optimization of large-scale systems, such as power grids, these results are potentially useful for other control and optimization applications as they shed a new light on both anti-windup control and projected gradient systems.