Search

Large Language Models (LLMs) can perform various natural language processing tasks with suitable instruction prompts. However, designing effective prompts manually is challenging and time-consuming. Existing methods for automatic prompt optimization either lack flexibility or efficiency. In this paper, we propose an effective approach to automatically select the optimal prompt for a given input from a finite set of synthetic candidate prompts. Our approach consists of three steps: (1) clustering the training data and generating candidate prompts for each cluster using an LLM-based prompt generator; (2) synthesizing a dataset of input-prompt-output tuples for training a prompt evaluator to rank the prompts based on their relevance to the input; (3) using the prompt evaluator to select the best prompt for a new input at test time. Our approach balances prompt generality-specificity and eliminates the need for resource-intensive training and inference. It demonstrates competitive performance on zero-shot question-answering datasets: GSM8K, MultiArith, and AQuA., Comment: preprint

In modern decentralized applications, ensuring communication efficiency and privacy for the users are the key challenges. In order to train machine-learning models, the algorithm has to communicate to the data center and sample data for its gradient computation, thus exposing the data and increasing the communication cost. This gives rise to the need for a decentralized optimization algorithm that is communication-efficient and minimizes the number of gradient computations. To this end, we propose the primal-dual sliding with conditional gradient sliding framework, which is communication-efficient and achieves an $\varepsilon$-approximate solution with the optimal gradient complexity of $O(1/\sqrt{\varepsilon}+\sigma^2/{\varepsilon^2})$ and $O(\log(1/\varepsilon)+\sigma^2/\varepsilon)$ for the convex and strongly convex setting respectively and an LO (Linear Optimization) complexity of $O(1/\varepsilon^2)$ for both settings given a stochastic gradient oracle with variance $\sigma^2$. Compared with the prior work \cite{wai-fw-2017}, our framework relaxes the assumption of the optimal solution being a strict interior point of the feasible set and enjoys wider applicability for large-scale training using a stochastic gradient oracle. We also demonstrate the efficiency of our algorithms with various numerical experiments.

Collaborative machine learning involves training models on data from multiple parties but must incentivize their participation. Existing data valuation methods fairly value and reward each party based on shared data or model parameters but neglect the privacy risks involved. To address this, we introduce differential privacy (DP) as an incentive. Each party can select its required DP guarantee and perturb its sufficient statistic (SS) accordingly. The mediator values the perturbed SS by the Bayesian surprise it elicits about the model parameters. As our valuation function enforces a privacy-valuation trade-off, parties are deterred from selecting excessive DP guarantees that reduce the utility of the grand coalition's model. Finally, the mediator rewards each party with different posterior samples of the model parameters. Such rewards still satisfy existing incentives like fairness but additionally preserve DP and a high similarity to the grand coalition's posterior. We empirically demonstrate the effectiveness and practicality of our approach on synthetic and real-world datasets., Comment: Accepted to NeurIPS 2023

This paper presents a simple yet efficient ensemble learning framework for Vietnamese scene text spotting. Leveraging the power of ensemble learning, which combines multiple models to yield more accurate predictions, our approach aims to significantly enhance the performance of scene text spotting in challenging urban settings. Through experimental evaluations on the VinText dataset, our proposed method achieves a significant improvement in accuracy compared to existing methods with an impressive accuracy of 5%. These results unequivocally demonstrate the efficacy of ensemble learning in the context of Vietnamese scene text spotting in urban environments, highlighting its potential for real world applications, such as text detection and recognition in urban signage, advertisements, and various text-rich urban scenes., Comment: RIVF 2023

This paper considers a class of noncoercive nonlinear elliptic problems with coefficients defined in Marcinkiewicz and Lorentz spaces. We prove the existence of a solution for the corresponding Dirichlet problem and investigate the higher integrability properties of the solution., Comment: 13 pages

We provide a concrete example exhibiting marked deviation from the PPN approximation in a modified theory of gravity. Specifically, we derive the exact formula for the Robertson parameter $\gamma$ in Brans-Dicke gravity for compact mass sources, explicitly incorporating the pressure content of these sources. We achieve this by exploiting the $\textit integrability$ of the 00-component of the Brans-Dicke field equation. In place of the conventional PPN result $\gamma_{PPN}=\frac{\omega+1}{\omega+2}$, we obtain the analytical expression $\gamma_{\,exact}=\frac{\omega+1+(\omega+2)\varTheta}{\omega+2+(\omega+1)\varTheta}$ where $\varTheta$ is the ratio of the total pressure $P_\parallel^*+2P_\perp^*$ and total energy $E^*$ contained within the mass source. Our $\textit non\text{-}perturbative$ formula is valid for all field strengths and types of matter comprising the mass source. We draw four key conclusions: (1) The usual $\gamma_{PPN}$ formula is violated in the presence of pressure, viz. when $\varTheta\neq0$, revealing a limitation of the PPN approximation in Brans-Dicke gravity. (2) The PPN result mainly stems from the assumption of pressureless matter. Even in the weak-field star case, non-zero pressure leads to a violation of the PPN $\gamma$ formula. Conversely, the PPN result is a good approximation for low-pressure matter, i.e. when $\varTheta\approx0$, for all field strengths. (3) Observational constraints on $\gamma$ set $\textit joint$ bounds on $\omega$ and $\varTheta$, with the latter representing a global characteristic of a mass source. If the equation of state of matter in the mass source approaches the ultra-relativistic form, entailing $\varTheta\simeq1$, $\gamma_{\,exact}$ converges to 1 $\textit irrespective$ of $\omega$. (4) In a broader context, our findings indicate the latent significance of considering the interior structure of stars in observational astronomy., Comment: To appear in EPJC. Figure 2 improved with new details; Section 9 added; Section 10 significantly expanded and enhanced

Using starlight polarization, we present the properties of foreground dust towards cluster NGC 7380 embedded in H{\sc ii} region Sh 2-142. Observations of starlight polarization are carried out in four filters using an imaging polarimeter equipped with a 104-cm ARIES telescope. Polarization vectors of stars are aligned along the Galactic magnetic field. Towards the east and southeast regions, the dust structure appears much denser than in other regions (inferred from extinction contours and colour composite image) and is also reflected in polarization distribution. We find that the polarization degree and extinction tend to increase with distance and indication for the presence of a dust layer at a distance of around 1.2 $kpc$. We have identified eight potential candidates exhibiting intrinsic polarization by employing three distinct criteria to distinguish between stars of intrinsic polarization and interstellar polarized stars. For interstellar polarized stars, we find that the maximum polarization degree increases with the color excess and has a strong scatter, with the mean value of 1.71$\pm$0.57$\%$. The peak wavelength spans $0.40-0.88\mu$m with the mean value of 0.56$\pm$0.07 $\mu m$, suggesting similar grain sizes in the region as the average diffuse interstellar medium. The polarization efficiency is also found to decrease with visual extinction as $P_{max}/A_{V}\propto A_{V}^{-0.61}$. Our observational results are found to be consistent with the predictions by the radiative torque alignment theory., Comment: 20 pages, 7 figures, 3 tables. Accepted for publication in AJ

In this work, we provide data stream algorithms that compute optimal splits in decision tree learning. In particular, given a data stream of observations $x_i$ and their labels $y_i$, the goal is to find the optimal split $j$ that divides the data into two sets such that the mean squared error (for regression) or misclassification rate and Gini impurity (for classification) is minimized. We provide several fast streaming algorithms that use sublinear space and a small number of passes for these problems. These algorithms can also be extended to the massively parallel computation model. Our work, while not directly comparable, complements the seminal work of Domingos-Hulten (KDD 2000) and Hulten-Spencer-Domingos (KDD 2001).

In the context of the model-driven development of data-centric applications, OCL constraints play a major role in adding precision to the source models (e.g., data models and security models). Several code-generators have been proposed to bridge the gap between source models with OCL constraints and their corresponding database implementations. However, the database queries produced by these code-generators are significantly less efficient -- from the point of view of execution-time performance -- than the implementations manually written by database experts. In this paper, we propose a different approach to bridge the gap between models with OCL constraints and their corresponding database implementations. In particular, we introduce a model-based methodology for proving the correctness of manually written SQL implementations of OCL constraints. This methodology is based on a novel mapping from a significant subset of the SQL language into many-sorted first-order logic. Moreover, by leveraging on an already existing mapping from the OCL language into many-sorted first-order logic, we can use SMT solvers to automatically prove the correctness of SQL implementations of OCL constraints. To illustrate and show the applicability of our approach, we include in the paper a number of non-trivial examples. Finally, we report on the status of a suite of tools supporting our approach., Comment: 11 pages

The rise of new video modalities like virtual reality or autonomous driving has increased the demand for efficient multi-view video compression methods, both in terms of rate-distortion (R-D) performance and in terms of delay and runtime. While most recent stereo video compression approaches have shown promising performance, they compress left and right views sequentially, leading to poor parallelization and runtime performance. This work presents Low-Latency neural codec for Stereo video Streaming (LLSS), a novel parallel stereo video coding method designed for fast and efficient low-latency stereo video streaming. Instead of using a sequential cross-view motion compensation like existing methods, LLSS introduces a bidirectional feature shifting module to directly exploit mutual information among views and encode them effectively with a joint cross-view prior model for entropy coding. Thanks to this design, LLSS processes left and right views in parallel, minimizing latency; all while substantially improving R-D performance compared to both existing neural and conventional codecs., Comment: Accepted by CVPR2024

Secure aggregation (SecAgg) is a commonly-used privacy-enhancing mechanism in federated learning, affording the server access only to the aggregate of model updates while safeguarding the confidentiality of individual updates. Despite widespread claims regarding SecAgg's privacy-preserving capabilities, a formal analysis of its privacy is lacking, making such presumptions unjustified. In this paper, we delve into the privacy implications of SecAgg by treating it as a local differential privacy (LDP) mechanism for each local update. We design a simple attack wherein an adversarial server seeks to discern which update vector a client submitted, out of two possible ones, in a single training round of federated learning under SecAgg. By conducting privacy auditing, we assess the success probability of this attack and quantify the LDP guarantees provided by SecAgg. Our numerical results unveil that, contrary to prevailing claims, SecAgg offers weak privacy against membership inference attacks even in a single training round. Indeed, it is difficult to hide a local update by adding other independent local updates when the updates are of high dimension. Our findings underscore the imperative for additional privacy-enhancing mechanisms, such as noise injection, in federated learning., Comment: accepted to the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD) 2024

Dust grains are important in various astrophysical processes and serve as indicators of interstellar medium structures, density, and mass. Understanding their physical properties and chemical composition is crucial in astrophysics. Dust polarisation is a valuable tool for studying these properties. The Radiative Torque (RAT) paradigm, which includes Radiative Torque Alignment (RAT-A) and Radiative Torque Disruption (RAT-D), is essential to interpret the dust polarisation data and constrain the fundamental properties of dust grains. However, it has been used primarily to interpret observations at a single wavelength. In this study, we analyse the thermal dust polarisation spectrum obtained from observations with SOFIA/HAWC+ and JCMT/POL-2 in the OMC-1 region and compare the observational data with our numerical results using the RAT paradigm. In general, we show that the dense gas exhibits a positive spectral slope, whereas the warm regions show a negative one. We demonstrate that a one-layer dust (one-phase) model can only reproduce the observed spectra at certain locations and cannot match those with prominent V-shaped spectra (for which the degree of polarisation initially decreases with wavelength from 54 to $\sim$ 300$\,\mu$m and then increases at longer wavelengths). To address this, we improve our model by incorporating two dust components (warm and cold) along the line of sight, resulting in a two-phase model. This improved model successfully reproduces the V-shaped spectra. The best model corresponds to a mixture composition of silicate and carbonaceous grains in the cold medium. Finally, by assuming the plausible model of grain alignment, we infer the inclination angle of the magnetic fields in OMC-1. This approach represents an important step towards better understanding the physics of grain alignment and constraining 3D magnetic fields using dust polarisation spectra., Comment: 12 pages, 9 figures, 3 tables, accepted to A&A

Polarization of starlight and thermal dust emission caused by aligned dust grains is a valuable tool to characterize magnetic fields (B-fields) and constrain dust properties. However, the physics of grain alignment is not fully understood. To test the popular paradigm of radiative torque (RAT) theory, including RAT alignment (RAT-A) and disruption (RAT-D), we use dust polarization data observed by {\it Planck} and SOFIA/HAWC+ toward two filaments with contrasting physical conditions: Musca, a quiet filament, and OMC-1, a highly dynamic filament due to feedback. We analyze various relations of the observed polarization fraction, $P$, with gas column density, $\NHt$, dust temperature, $\Td$, and polarization angle dispersion function, $\S$. We found that $P$ decreases with increasing $\S$ and increasing $\NHt$, as expected from RAT-A. On the other hand, the $P-\Td$ relation is more complicated; it is a linear correlation at low $\Td$ but turns into an anti-correlation when $\Td$ reaches a certain high value. Next, we compute the polarization fraction on a pixel-by-pixel with B-fields in the plane of the sky using the DustPOL code based on RAT, incorporate the depolarization effect by B-field tangling using $\S$, and compare the realistic polarization model with observations of Musca and OMC-1. For Musca with well-ordered B-fields, our numerical model reproduces the decline of $P$ toward the filament spine (aka. polarization hole), having high $\NHt$ and low $\Td$, indicating the loss of grain alignment efficiency due to RAT-A. For OMC-1, with stronger B-field variations and higher $\Td$, our model can reproduce the observed $P-\Td$ and $P-N(\rm H_{2})$ relations only if the depolarization effect resulting from B-field tangling and RAT-D effect are taken into account. Our results provide more robust observational evidence for the RAT paradigm, particularly the recently discovered RAT-D., Comment: 13 figures, 24 pages

The proliferation of online toxic speech is a pertinent problem posing threats to demographic groups. While explicit toxic speech contains offensive lexical signals, implicit one consists of coded or indirect language. Therefore, it is crucial for models not only to detect implicit toxic speech but also to explain its toxicity. This draws a unique need for unified frameworks that can effectively detect and explain implicit toxic speech. Prior works mainly formulated the task of toxic speech detection and explanation as a text generation problem. Nonetheless, models trained using this strategy can be prone to suffer from the consequent error propagation problem. Moreover, our experiments reveal that the detection results of such models are much lower than those that focus only on the detection task. To bridge these gaps, we introduce ToXCL, a unified framework for the detection and explanation of implicit toxic speech. Our model consists of three modules: a (i) Target Group Generator to generate the targeted demographic group(s) of a given post; an (ii) Encoder-Decoder Model in which the encoder focuses on detecting implicit toxic speech and is boosted by a (iii) Teacher Classifier via knowledge distillation, and the decoder generates the necessary explanation. ToXCL achieves new state-of-the-art effectiveness, and outperforms baselines significantly., Comment: Accepted at NAACL 2024 (Main Conference)

Personas are commonly used in software projects to gain a better understanding of end-users' needs. However, there is a limited understanding of their usage and effectiveness in practice. This paper presents the results of a two-step investigation, comprising interviews with 26 software developers, UI/UX designers, business analysts and product managers and a survey of 203 practitioners, aimed at shedding light on the current practices, methods and challenges of using personas in software development. Our findings reveal variations in the frequency and effectiveness of personas across different software projects and IT companies, the challenges practitioners face when using personas and the reasons for not using them at all. Furthermore, we investigate the coverage of human aspects in personas, often assumed to be a key feature of persona descriptions. Contrary to the general perception, our study shows that human aspects are often ignored for various reasons in personas or requirements engineering in general. Our study provides actionable insights for practitioners to overcome challenges in using personas during requirements engineering stages, and we identify areas for future research.

The success of Natural Language Understanding (NLU) benchmarks in various languages, such as GLUE for English, CLUE for Chinese, KLUE for Korean, and IndoNLU for Indonesian, has facilitated the evaluation of new NLU models across a wide range of tasks. To establish a standardized set of benchmarks for Vietnamese NLU, we introduce the first Vietnamese Language Understanding Evaluation (VLUE) benchmark. The VLUE benchmark encompasses five datasets covering different NLU tasks, including text classification, span extraction, and natural language understanding. To provide an insightful overview of the current state of Vietnamese NLU, we then evaluate seven state-of-the-art pre-trained models, including both multilingual and Vietnamese monolingual models, on our proposed VLUE benchmark. Furthermore, we present CafeBERT, a new state-of-the-art pre-trained model that achieves superior results across all tasks in the VLUE benchmark. Our model combines the proficiency of a multilingual pre-trained model with Vietnamese linguistic knowledge. CafeBERT is developed based on the XLM-RoBERTa model, with an additional pretraining step utilizing a significant amount of Vietnamese textual data to enhance its adaptation to the Vietnamese language. For the purpose of future research, CafeBERT is made publicly available for research purposes., Comment: Accepted at NAACL 2024 (Findings)

In cellular networks, authorities may need to physically locate user devices to track criminals or illegal equipment. This process involves authorized agents tracing devices by monitoring uplink signals with cellular operator assistance. However, tracking uncooperative uplink signal sources remains challenging, even for operators and authorities. Three key challenges persist for fine-grained localization: i) devices must generate sufficient, consistent uplink traffic over time, ii) target devices may transmit uplink signals at very low power, and iii) signals from cellular repeaters may hinder localization of the target device. While these challenges pose significant practical obstacles to localization, they have been largely overlooked in existing research. This work examines the impact of these real-world challenges on cellular localization and introduces the Uncooperative Multiangulation Attack (UMA) to address them. UMA can 1) force a target device to transmit traffic continuously, 2) boost the target's signal strength to maximum levels, and 3) uniquely differentiate between signals from the target and repeaters. Importantly, UMA operates without requiring privileged access to cellular operators or user devices, making it applicable to any LTE network. Our evaluations demonstrate that UMA effectively overcomes practical challenges in physical localization when devices are uncooperative.

While intrusion detection systems (IDSs) benefit from the diversity and generalization of IoT data features, the data diversity (e.g., the heterogeneity and high dimensions of data) also makes it difficult to train effective machine learning models in IoT IDSs. This also leads to potentially redundant/noisy features that may decrease the accuracy of the detection engine in IDSs. This paper first introduces a novel neural network architecture called Multiple-Input Auto-Encoder (MIAE). MIAE consists of multiple sub-encoders that can process inputs from different sources with different characteristics. The MIAE model is trained in an unsupervised learning mode to transform the heterogeneous inputs into lower-dimensional representation, which helps classifiers distinguish between normal behaviour and different types of attacks. To distil and retain more relevant features but remove less important/redundant ones during the training process, we further design and embed a feature selection layer right after the representation layer of MIAE resulting in a new model called MIAEFS. This layer learns the importance of features in the representation vector, facilitating the selection of informative features from the representation vector. The results on three IDS datasets, i.e., NSLKDD, UNSW-NB15, and IDS2017, show the superior performance of MIAE and MIAEFS compared to other methods, e.g., conventional classifiers, dimensionality reduction models, unsupervised representation learning methods with different input dimensions, and unsupervised feature selection models. Moreover, MIAE and MIAEFS combined with the Random Forest (RF) classifier achieve accuracy of 96.5% in detecting sophisticated attacks, e.g., Slowloris. The average running time for detecting an attack sample using RF with the representation of MIAE and MIAEFS is approximate 1.7E-6 seconds, whilst the model size is lower than 1 MB.

Representation Learning (RL) plays a pivotal role in the success of many problems including cyberattack detection. Most of the RL methods for cyberattack detection are based on the latent vector of Auto-Encoder (AE) models. An AE transforms raw data into a new latent representation that better exposes the underlying characteristics of the input data. Thus, it is very useful for identifying cyberattacks. However, due to the heterogeneity and sophistication of cyberattacks, the representation of AEs is often entangled/mixed resulting in the difficulty for downstream attack detection models. To tackle this problem, we propose a novel mod called Twin Auto-Encoder (TAE). TAE deterministically transforms the latent representation into a more distinguishable representation namely the \textit{separable representation} and the reconstructsuct the separable representation at the output. The output of TAE called the \textit{reconstruction representation} is input to downstream models to detect cyberattacks. We extensively evaluate the effectiveness of TAE using a wide range of bench-marking datasets. Experiment results show the superior accuracy of TAE over state-of-the-art RL models and well-known machine learning algorithms. Moreover, TAE also outperforms state-of-the-art models on some sophisticated and challenging attacks. We then investigate various characteristics of TAE to further demonstrate its superiority.

We propose a distributional formulation of the spanning problem of a multi-asset payoff by vanilla basket options. This problem is shown to have a unique solution if and only if the payoff function is even and absolutely homogeneous, and we establish a Fourier-based formula to calculate the solution. Financial payoffs are typically piecewise linear, resulting in a solution that may be derived explicitly, yet may also be hard to numerically exploit. One-hidden-layer feedforward neural networks instead provide a natural and efficient numerical alternative for discrete spanning. We test this approach for a selection of archetypal payoffs and obtain better hedging results with vanilla basket options compared to industry-favored approaches based on single-asset vanilla hedges.

In the realm of Federated Learning (FL) applied to remote sensing image classification, this study introduces and assesses several innovative communication strategies. Our exploration includes feature-centric communication, pseudo-weight amalgamation, and a combined method utilizing both weights and features. Experiments conducted on two public scene classification datasets unveil the effectiveness of these strategies, showcasing accelerated convergence, heightened privacy, and reduced network information exchange. This research provides valuable insights into the implications of feature-centric communication in FL, offering potential applications tailored for remote sensing scenarios., Comment: 5 pages, to appear in IGARSS 2024

This paper focuses on estimating the invariant density function $f_X$ of the strongly mixing stationary process $X_t$ in the multiplicative measurement errors model $Y_t = X_t U_t$, where $U_t$ is also a strongly mixing stationary process. We propose a novel approach to handle non-independent data, typical in real-world scenarios. For instance, data collected from various groups may exhibit interdependencies within each group, resembling data generated from $m$-dependent stationary processes, a subset of stationary processes. This study extends the applicability of the model $Y_t = X_t U_t$ to diverse scientific domains dealing with complex dependent data. The paper outlines our estimation techniques, discusses convergence rates, establishes a lower bound on the minimax risk, and demonstrates the asymptotic normality of the estimator for $f_X$ under smooth error distributions. Through examples and simulations, we showcase the efficacy of our estimator. The paper concludes by providing proofs for the presented theoretical results.v

Uncertainty in medical image segmentation tasks, especially inter-rater variability, arising from differences in interpretations and annotations by various experts, presents a significant challenge in achieving consistent and reliable image segmentation. This variability not only reflects the inherent complexity and subjective nature of medical image interpretation but also directly impacts the development and evaluation of automated segmentation algorithms. Accurately modeling and quantifying this variability is essential for enhancing the robustness and clinical applicability of these algorithms. We report the set-up and summarize the benchmark results of the Quantification of Uncertainties in Biomedical Image Quantification Challenge (QUBIQ), which was organized in conjunction with International Conferences on Medical Image Computing and Computer-Assisted Intervention (MICCAI) 2020 and 2021. The challenge focuses on the uncertainty quantification of medical image segmentation which considers the omnipresence of inter-rater variability in imaging datasets. The large collection of images with multi-rater annotations features various modalities such as MRI and CT; various organs such as the brain, prostate, kidney, and pancreas; and different image dimensions 2D-vs-3D. A total of 24 teams submitted different solutions to the problem, combining various baseline models, Bayesian neural networks, and ensemble model techniques. The obtained results indicate the importance of the ensemble models, as well as the need for further research to develop efficient 3D methods for uncertainty quantification methods in 3D segmentation tasks., Comment: initial technical report

Facial Expression Recognition (FER) is a critical task within computer vision with diverse applications across various domains. Addressing the challenge of limited FER datasets, which hampers the generalization capability of expression recognition models, is imperative for enhancing performance. Our paper presents an innovative approach integrating the MAE-Face self-supervised learning (SSL) method and multi-view Fusion Attention mechanism for expression classification, particularly showcased in the 6th Affective Behavior Analysis in-the-wild (ABAW) competition. By utilizing low-level feature information from the ipsilateral view (auxiliary view) before learning the high-level feature that emphasizes the shift in the human facial expression, our work seeks to provide a straightforward yet innovative way to improve the examined view (main view). We also suggest easy-to-implement and no-training frameworks aimed at highlighting key facial features to determine if such features can serve as guides for the model, focusing on pivotal local elements. The efficacy of this method is validated by improvements in model performance on the Aff-wild2 dataset, as observed in both training and validation contexts., Comment: 6 pages; added references for section 1; corrected typo for email author

Perception plays a crucial role in various robot applications. However, existing well-annotated datasets are biased towards autonomous driving scenarios, while unlabelled SLAM datasets are quickly over-fitted, and often lack environment and domain variations. To expand the frontier of these fields, we introduce a comprehensive dataset named MCD (Multi-Campus Dataset), featuring a wide range of sensing modalities, high-accuracy ground truth, and diverse challenging environments across three Eurasian university campuses. MCD comprises both CCS (Classical Cylindrical Spinning) and NRE (Non-Repetitive Epicyclic) lidars, high-quality IMUs (Inertial Measurement Units), cameras, and UWB (Ultra-WideBand) sensors. Furthermore, in a pioneering effort, we introduce semantic annotations of 29 classes over 59k sparse NRE lidar scans across three domains, thus providing a novel challenge to existing semantic segmentation research upon this largely unexplored lidar modality. Finally, we propose, for the first time to the best of our knowledge, continuous-time ground truth based on optimization-based registration of lidar-inertial data on large survey-grade prior maps, which are also publicly released, each several times the size of existing ones. We conduct a rigorous evaluation of numerous state-of-the-art algorithms on MCD, report their performance, and highlight the challenges awaiting solutions from the research community., Comment: Accepted by The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2024

This research paper presents an approach to enhancing the predictive capability of architects in the design and assurance of systems, focusing on systems operating in dynamic and unpredictable environments. By adopting a systems approach, we aim to improve architects' predictive capabilities in designing dependable systems (for example, ML-based systems). An aerospace case study is used to illustrate the approach. Multiple factors (challenges) influencing aircraft detection are identified, demonstrating the effectiveness of our approach in a complex operational setting. Our approach primarily aimed to enhance the architect's predictive capability., Comment: S. Bernardi, T. Zoppi (Editors), "Fast Abstracts and Student Forum Proceedings - EDCC 2024 - 19th European Dependable Computing Conference, Leuven, Belgium, 8-11 April 2024"

Classical structural-based visual localization methods offer high accuracy but face trade-offs in terms of storage, speed, and privacy. A recent innovation, keypoint scene coordinate regression (KSCR) named D2S addresses these issues by leveraging graph attention networks to enhance keypoint relationships and predict their 3D coordinates using a simple multilayer perceptron (MLP). Camera pose is then determined via PnP+RANSAC, using established 2D-3D correspondences. While KSCR achieves competitive results, rivaling state-of-the-art image-retrieval methods like HLoc across multiple benchmarks, its performance is hindered when data samples are limited due to the deep learning model's reliance on extensive data. This paper proposes a solution to this challenge by introducing a pipeline for keypoint descriptor synthesis using Neural Radiance Field (NeRF). By generating novel poses and feeding them into a trained NeRF model to create new views, our approach enhances the KSCR's generalization capabilities in data-scarce environments. The proposed system could significantly improve localization accuracy by up to 50% and cost only a fraction of time for data synthesis. Furthermore, its modular design allows for the integration of multiple NeRFs, offering a versatile and efficient solution for visual localization. The implementation is publicly available at: https://github.com/ais-lab/DescriptorSynthesis4Feat2Map.

Drawing is an art that enables people to express their imagination and emotions. However, individuals usually face challenges in drawing, especially when translating conceptual ideas into visually coherent representations and bridging the gap between mental visualization and practical execution. In response, we propose ARtVista - a novel system integrating AR and generative AI technologies. ARtVista not only recommends reference images aligned with users' abstract ideas and generates sketches for users to draw but also goes beyond, crafting vibrant paintings in various painting styles. ARtVista also offers users an alternative approach to create striking paintings by simulating the paint-by-number concept on reference images, empowering users to create visually stunning artwork devoid of the necessity for advanced drawing skills. We perform a pilot study and reveal positive feedback on its usability, emphasizing its effectiveness in visualizing user ideas and aiding the painting process to achieve stunning pictures without requiring advanced drawing skills. The source code will be available at https://github.com/htrvu/ARtVista., Comment: CHI 2024

Resonant Inelastic X-ray Scattering (RIXS) is an ideal X-ray spectroscopy method to push the combination of energy and time resolutions to the Fourier transform ultimate limit, because it is unaffected by the core-hole lifetime energy broadening. And in pump-probe experiments the interaction time is made very short by the same core-hole lifetime. RIXS is very photon hungry so it takes great advantage from high repetition rate pulsed X-ray sources like the European XFEL. The hRIXS instrument is designed for RIXS experiments in the soft X-ray range with energy resolution approaching the Fourier and the Heisenberg limits. It is based on a spherical grating with variable line spacing (VLS) and a position-sensitive 2D detector. Initially, two gratings are installed to adequately cover the whole photon energy range. With optimized spot size on the sample and small pixel detector the energy resolution can be better than 40 meV at any photon energy below 1000 eV. At the SCS instrument of the European XFEL the spectrometer can be easily positioned thanks to air-pads on a high-quality floor, allowing the scattering angle to be continuously adjusted over the 65-145 deg range. It can be coupled to two different sample interaction chamber, one for liquid jets and one for solids, each equipped at the state-of-the-art and compatible for optical laser pumping in collinear geometry. The measured performances, in terms of energy resolution and count rate on the detector, closely match design expectations. hRIXS is open to public users since the summer of 2022., Comment: 43 pages, 12 figures, Supplemental Material

Graph Neural Networks (GNNs), known as spectral graph filters, find a wide range of applications in web networks. To bypass eigendecomposition, polynomial graph filters are proposed to approximate graph filters by leveraging various polynomial bases for filter training. However, no existing studies have explored the diverse polynomial graph filters from a unified perspective for optimization. In this paper, we first unify polynomial graph filters, as well as the optimal filters of identical degrees into the Krylov subspace of the same order, thus providing equivalent expressive power theoretically. Next, we investigate the asymptotic convergence property of polynomials from the unified Krylov subspace perspective, revealing their limited adaptability in graphs with varying heterophily degrees. Inspired by those facts, we design a novel adaptive Krylov subspace approach to optimize polynomial bases with provable controllability over the graph spectrum so as to adapt various heterophily graphs. Subsequently, we propose AdaptKry, an optimized polynomial graph filter utilizing bases from the adaptive Krylov subspaces. Meanwhile, in light of the diverse spectral properties of complex graphs, we extend AdaptKry by leveraging multiple adaptive Krylov bases without incurring extra training costs. As a consequence, extended AdaptKry is able to capture the intricate characteristics of graphs and provide insights into their inherent complexity. We conduct extensive experiments across a series of real-world datasets. The experimental results demonstrate the superior filtering capability of AdaptKry, as well as the optimized efficacy of the adaptive Krylov basis.

Terrestrial networks form the fundamental infrastructure of modern communication systems, serving more than 4 billion users globally. However, terrestrial networks are facing a wide range of challenges, from coverage and reliability to interference and congestion. As the demands of the 6G era are expected to be much higher, it is crucial to address these challenges to ensure a robust and efficient communication infrastructure for the future. To address these problems, Non-terrestrial Network (NTN) has emerged to be a promising solution. NTNs are communication networks that leverage airborne (e.g., unmanned aerial vehicles) and spaceborne vehicles (e.g., satellites) to facilitate ultra-reliable communications and connectivity with high data rates and low latency over expansive regions. This article aims to provide a comprehensive survey on the utilization of network slicing, Artificial Intelligence/Machine Learning (AI/ML), and Open Radio Access Network (ORAN) to address diverse challenges of NTNs from the perspectives of both academia and industry. Particularly, we first provide an in-depth tutorial on NTN and the key enabling technologies including network slicing, AI/ML, and ORAN. Then, we provide a comprehensive survey on how network slicing and AI/ML have been leveraged to overcome the challenges that NTNs are facing. Moreover, we present how ORAN can be utilized for NTNs. Finally, we highlight important challenges, open issues, and future research directions of NTN in the 6G era., Comment: 35 pages

We present a fully autonomous self-recharging drone system capable of long-duration sustained operations near powerlines. The drone is equipped with a robust onboard perception and navigation system that enables it to locate powerlines and approach them for landing. A passively actuated gripping mechanism grasps the powerline cable during landing after which a control circuit regulates the magnetic field inside a split-core current transformer to provide sufficient holding force as well as battery recharging. The system is evaluated in an active outdoor three-phase powerline environment. We demonstrate multiple contiguous hours of fully autonomous uninterrupted drone operations composed of several cycles of flying, landing, recharging, and takeoff, validating the capability of extended, essentially unlimited, operational endurance., Comment: Accepted for publication at the 2024 IEEE International Conference on Robotics and Automation (ICRA). Video at https://www.youtube.com/watch?v=C-uekD6VTIQ

Code Large Language Models (CodeLLMs) have demonstrated impressive proficiency in code completion tasks. However, they often fall short of fully understanding the extensive context of a project repository, such as the intricacies of relevant files and class hierarchies, which can result in less precise completions. To overcome these limitations, we present \tool, a multifaceted framework designed to address the complex challenges associated with repository-level code completion. Central to RepoHYPER is the {\em Repo-level Semantic Graph} (RSG), a novel semantic graph structure that encapsulates the vast context of code repositories. Furthermore, RepoHyper leverages Expand and Refine retrieval method, including a graph expansion and a link prediction algorithm applied to the RSG, enabling the effective retrieval and prioritization of relevant code snippets. Our evaluations show that \tool markedly outperforms existing techniques in repository-level code completion, showcasing enhanced accuracy across various datasets when compared to several strong baselines. Our implementation of RepoHYPER can be found at https://github.com/FSoft-AI4Code/RepoHyper.

In this report, we introduce the Gemini 1.5 family of models, representing the next generation of highly compute-efficient multimodal models capable of recalling and reasoning over fine-grained information from millions of tokens of context, including multiple long documents and hours of video and audio. The family includes two new models: (1) an updated Gemini 1.5 Pro, which exceeds the February version on the great majority of capabilities and benchmarks; (2) Gemini 1.5 Flash, a more lightweight variant designed for efficiency with minimal regression in quality. Gemini 1.5 models achieve near-perfect recall on long-context retrieval tasks across modalities, improve the state-of-the-art in long-document QA, long-video QA and long-context ASR, and match or surpass Gemini 1.0 Ultra's state-of-the-art performance across a broad set of benchmarks. Studying the limits of Gemini 1.5's long-context ability, we find continued improvement in next-token prediction and near-perfect retrieval (>99%) up to at least 10M tokens, a generational leap over existing models such as Claude 3.0 (200k) and GPT-4 Turbo (128k). Finally, we highlight real-world use cases, such as Gemini 1.5 collaborating with professionals on completing their tasks achieving 26 to 75% time savings across 10 different job categories, as well as surprising new capabilities of large language models at the frontier; when given a grammar manual for Kalamang, a language with fewer than 200 speakers worldwide, the model learns to translate English to Kalamang at a similar level to a person who learned from the same content.

In this paper, we propose a radar odometry structure that directly utilizes radar velocity measurements for dead reckoning while maintaining its ability to update estimations within the Kalman filter framework. Specifically, we employ the Doppler velocity obtained by a 4D Frequency Modulated Continuous Wave (FMCW) radar in conjunction with gyroscope data to calculate poses. This approach helps mitigate high drift resulting from accelerometer biases and double integration. Instead, tilt angles measured by gravitational force are utilized alongside relative distance measurements from radar scan matching for the filter's measurement update. Additionally, to further enhance the system's accuracy, we estimate and compensate for the radar velocity scale factor. The performance of the proposed method is verified through five real-world open-source datasets. The results demonstrate that our approach reduces position error by 62% and rotation error by 66% on average compared to the state-of-the-art radar-inertial fusion method in terms of absolute trajectory error., Comment: 9 pages, 5 figures, 1 table, IROS 2024

Two-dimensional (2D) semiconducting transition-metal dichalcogenides (TMDCs) are an exciting platform for new excitonic physics and next-generation electronics, creating a strong demand to understand their growth, doping, and heterostructures. Despite significant progress in solid-source (SS-) and metal-organic chemical vapor deposition (MOCVD), further optimization is necessary to grow highly crystalline 2D TMDCs with controlled doping. Here, we report a hybrid MOCVD growth method that combines liquid-phase metal precursor deposition and vapor-phase organo-chalcogen delivery to leverage the advantages of both MOCVD and SS-CVD. Using our hybrid approach, we demonstrate WS$_2$ growth with tunable morphologies - from separated single-crystal domains to continuous monolayer films - on a variety of substrates, including sapphire, SiO$_2$, and Au. These WS$_2$ films exhibit narrow neutral exciton photoluminescence linewidths down to 33 meV and room-temperature mobility up to 34 - 36 cm$^2$V$^-$$^1$s$^-$$^1$). Through simple modifications to the liquid precursor composition, we demonstrate the growth of V-doped WS$_2$, MoxW$_1$$_-$$_x$S$_2$ alloys, and in-plane WS$_2$-MoS$_2$ heterostructures. This work presents an efficient approach for addressing a variety of TMDC synthesis needs on a laboratory scale.

Models trained on data composed of different groups or domains can suffer from severe performance degradation under distribution shifts. While recent methods have largely focused on optimizing the worst-group objective, this often comes at the expense of good performance on other groups. To address this problem, we introduce an optimization scheme to achieve good performance across groups and find a good solution for all without severely sacrificing performance on any of them. However, directly applying such optimization involves updating the parameters of the entire network, making it both computationally expensive and challenging. Thus, we introduce Controllable Prompt Tuning (CPT), which couples our approach with prompt-tuning techniques. On spurious correlation benchmarks, our procedures achieve state-of-the-art results across both transformer and non-transformer architectures, as well as unimodal and multimodal data, while requiring only 0.4% tunable parameters., Comment: Proceedings of the 41st International Conference on Machine Learning

In this new era of rapid AI development, especially in language processing, the demand for AI in the legal domain is increasingly critical. In the context where research in other languages such as English, Japanese, and Chinese has been well-established, we introduce the first fundamental research for the Vietnamese language in the legal domain: legal textual entailment recognition through the Vietnamese Language and Speech Processing workshop. In analyzing participants' results, we discuss certain linguistic aspects critical in the legal domain that pose challenges that need to be addressed.

We study the three-dimensional Navier-Stokes equations in a periodic domain with the force decaying in time. Although the force has a certain coherent decay, as time tends to infinity, it can be too complicated for the previous theory of asymptotic expansions to be applicable. To deal with this issue, we systematically develop a new theory of asymptotic expansions containing the so-called subordinate variables which can be defined recursively. We apply it to obtain an asymptotic expansion for any Leray-Hopf weak solutions. The expansion, in fact, is constructed explicitly and the impact of the subordinate variables can be clearly specified. The complexifications of the Gevrey-Sobolev spaces, and of the Stokes and bilinear operators of the Navier-Stokes equations are utilized to facilitate such a construction., Comment: 40 pp, submitted for publication. arXiv admin note: text overlap with arXiv:2204.05247

Among the various candidates for dark matter (DM), ultralight vector DM can be probed by laser interferometric gravitational wave detectors through the measurement of oscillating length changes in the arm cavities. In this context, KAGRA has a unique feature due to differing compositions of its mirrors, enhancing the signal of vector DM in the length change in the auxiliary channels. Here we present the result of a search for $U(1)_{B-L}$ gauge boson DM using the KAGRA data from auxiliary length channels during the first joint observation run together with GEO600. By applying our search pipeline, which takes into account the stochastic nature of ultralight DM, upper bounds on the coupling strength between the $U(1)_{B-L}$ gauge boson and ordinary matter are obtained for a range of DM masses. While our constraints are less stringent than those derived from previous experiments, this study demonstrates the applicability of our method to the lower-mass vector DM search, which is made difficult in this measurement by the short observation time compared to the auto-correlation time scale of DM., Comment: 20 pages, 5 figures

A timing and synchronization system based on a de Bruijn sequence has been proposed and studied recently for a channel associated with quantum communication that requires reliable synchronization. To avoid a long period of no-pulse in such a system on-off pulses are used to simulate a zero and on-on pulses are used to simulate a one. However, these sequences have high redundancy. To reduce the redundancy, run-length limited sequences in the de Bruijn graph are proposed for the same purpose. The maximum length of such sequences in the de Bruijn graph is studied and an efficient algorithm to construct a large set of these sequences is presented. A maximum length sequence for which the position of each window can be computed efficiently is constructed. Finally, an enumeration of the number of such sequences is given and some generalizations are discussed.

In federated learning, the models can be trained synchronously or asynchronously. Many research works have focused on developing an aggregation method for the server to aggregate multiple local models into the global model with improved performance. They ignore the heterogeneity of the training workers, which causes the delay in the training of the local models, leading to the obsolete information issue. In this paper, we design and develop Asyn2F, an Asynchronous Federated learning Framework with bidirectional model aggregation. By bidirectional model aggregation, Asyn2F, on one hand, allows the server to asynchronously aggregate multiple local models and results in a new global model. On the other hand, it allows the training workers to aggregate the new version of the global model into the local model, which is being trained even in the middle of a training epoch. We develop Asyn2F considering the practical implementation requirements such as using cloud services for model storage and message queuing protocols for communications. Extensive experiments with different datasets show that the models trained by Asyn2F achieve higher performance compared to the state-of-the-art techniques. The experiments also demonstrate the effectiveness, practicality, and scalability of Asyn2F, making it ready for deployment in real scenarios.

In this paper, we propose a cost-effective strategy for heterogeneous UAV swarm systems for cooperative aerial inspection. Unlike previous swarm inspection works, the proposed method does not rely on precise prior knowledge of the environment and can complete full 3D surface coverage of objects in any shape. In this work, agents are partitioned into teams, with each drone assign a different task, including mapping, exploration, and inspection. Task allocation is facilitated by assigning optimal inspection volumes to each team, following best-first rules. A voxel map-based representation of the environment is used for pathfinding, and a rule-based path-planning method is the core of this approach. We achieved the best performance in all challenging experiments with the proposed approach, surpassing all benchmark methods for similar tasks across multiple evaluation trials. The proposed method is open source at https://github.com/ntu-aris/caric_baseline and used as the baseline of the Cooperative Aerial Robots Inspection Challenge at the 62nd IEEE Conference on Decision and Control 2023., Comment: Baseline method of CARIC at CDC 2023, Singapore

This paper presents a distributed solution for the problem of collaborative collision avoidance for autonomous inland waterway ships. A two-layer collision avoidance framework that considers inland waterway traffic regulations is proposed to increase navigational safety for autonomous ships. Our approach allows for modifying traffic rules without changing the collision avoidance algorithm, and is based on a novel formulation of model predictive control (MPC) for collision avoidance of ships. This MPC formulation is designed for inland waterway traffic and can handle complex scenarios. The alternating direction method of multipliers is used as a scheme for exchanging and negotiating intentions among ships. Simulation results show that the proposed algorithm can comply with traffic rules. Furthermore, the proposed algorithm can safely deviate from traffic rules when necessary to increase efficiency in complex scenarios.

With recent advances in artificial intelligence (AI) and robotics, unmanned vehicle swarms have received great attention from both academia and industry due to their potential to provide services that are difficult and dangerous to perform by humans. However, learning and coordinating movements and actions for a large number of unmanned vehicles in complex and dynamic environments introduce significant challenges to conventional AI methods. Generative AI (GAI), with its capabilities in complex data feature extraction, transformation, and enhancement, offers great potential in solving these challenges of unmanned vehicle swarms. For that, this paper aims to provide a comprehensive survey on applications, challenges, and opportunities of GAI in unmanned vehicle swarms. Specifically, we first present an overview of unmanned vehicles and unmanned vehicle swarms as well as their use cases and existing issues. Then, an in-depth background of various GAI techniques together with their capabilities in enhancing unmanned vehicle swarms are provided. After that, we present a comprehensive review on the applications and challenges of GAI in unmanned vehicle swarms with various insights and discussions. Finally, we highlight open issues of GAI in unmanned vehicle swarms and discuss potential research directions., Comment: 23 pages

Recent advancements in visual localization and mapping have demonstrated considerable success in integrating point and line features. However, expanding the localization framework to include additional mapping components frequently results in increased demand for memory and computational resources dedicated to matching tasks. In this study, we show how a lightweight neural network can learn to represent both 3D point and line features, and exhibit leading pose accuracy by harnessing the power of multiple learned mappings. Specifically, we utilize a single transformer block to encode line features, effectively transforming them into distinctive point-like descriptors. Subsequently, we treat these point and line descriptor sets as distinct yet interconnected feature sets. Through the integration of self- and cross-attention within several graph layers, our method effectively refines each feature before regressing 3D maps using two simple MLPs. In comprehensive experiments, our indoor localization findings surpass those of Hloc and Limap across both point-based and line-assisted configurations. Moreover, in outdoor scenarios, our method secures a significant lead, marking the most considerable enhancement over state-of-the-art learning-based methodologies. The source code and demo videos of this work are publicly available at: https://thpjp.github.io/pl2map/

Radio jets are present in a diverse sample of AGN. However, the mechanisms of jet powering are not fully understood, and it is yet unclear to what extent they obey mass-invariant scaling relations, similar to those found for the triggering and fuelling of X-ray selected AGN. We study the incidence of eROSITA/eFEDS X-ray and LOFAR radio AGN as a function of several stellar mass normalised AGN power indicators. A new sample of radio AGN from the LOFAR-eFEDS survey is defined and we publicly release this catalogue, including host galaxy counterparts from the Legacy Survey DR9, LOFAR radio morphologies and host galaxy properties from the complete, spectroscopic (z<0.4) GAMA09 survey. The fraction of GAMA09 galaxies hosting radio, X-ray and both radio and X-ray AGN are calculated as a function of the specific black hole kinetic ($\lambda_{\rm Jet}$) and radiative ($\lambda_{\rm Edd}$) power. The incidence of eFEDS X-ray AGN as a function of $\lambda_{\rm Edd}$ shows the same mass-invariance as found in past studies. Meanwhile, radio AGN, regardless of their morphology, are more likely to be hosted in more massive galaxies, at all $\lambda_{\rm Jet}$. Across the stellar mass range, the compact radio AGN incidence follows the same power-law distribution, showing that it is not only high mass galaxies that host high power radio AGN and vice versa. On the other hand, the incidence of compact and complex radio AGN is boosted at the highest jet powers, diverging from a simple power-law. Interestingly, this increased incidence cannot be explained by more powerful radio AGN lying in more dense environments which could naturally boost their radio luminosity. Overall, we show that statistical incidence studies are a powerful method to probe disk-jet coupling for different AGN accretion modes, although future work on a more reliable determination of jet power for diverse samples of radio AGN is needed., Comment: 30 pages, 32 figures, accepted for publication in A&A

A novel class of Buchdahl-inspired metrics with closed-form expressions was recently obtained based on Buchdahl's seminal work on searching for static, spherically symmetric metrics in ${\cal R}^{2}$ gravity in vacuo. Buchdahl-inspired spacetimes provide an interesting framework for testing predictions of ${\cal R}^{2}$ gravity models against observations. To test these Buchdahl-inspired spacetimes, we consider observational constraints imposed on the deviation parameter, which characterizes the deviation of the asymptotically flat Buchdahl-inspired metric from the Schwarzschild spacetime. We utilize several recent solar system experiments and observations of the S2 star in the Galactic center and the black hole shadow. By calculating the effects of Buchdahl-inspired spacetimes on astronomical observations both within and outside of the solar system, including the deflection angle of light by the Sun, gravitational time delay, perihelion advance, shadow, and geodetic precession, we determine observational constraints on the corresponding deviation parameters by comparing theoretical predictions with the most recent observations. Among these constraints, we find that the tightest one comes from the Cassini mission's measurement of gravitational time delay., Comment: 13 pages, no figure, to appear in EPJC

Autonomous agents powered by large language models (LLMs) have garnered significant research attention. However, fully harnessing the potential of LLMs for agent-based tasks presents inherent challenges due to the heterogeneous nature of diverse data sources featuring multi-turn trajectories. In this paper, we introduce \textbf{AgentOhana} as a comprehensive solution to address these challenges. \textit{AgentOhana} aggregates agent trajectories from distinct environments, spanning a wide array of scenarios. It meticulously standardizes and unifies these trajectories into a consistent format, streamlining the creation of a generic data loader optimized for agent training. Leveraging the data unification, our training pipeline maintains equilibrium across different data sources and preserves independent randomness across devices during dataset partitioning and model training. Additionally, we present \textbf{xLAM-v0.1}, a large action model tailored for AI agents, which demonstrates exceptional performance across various benchmarks. Begin the exploration at \url{https://github.com/SalesforceAIResearch/xLAM}., Comment: Add GitHub repo link at \url{https://github.com/SalesforceAIResearch/xLAM} and HuggingFace model link at \url{https://huggingface.co/Salesforce/xLAM-v0.1-r}

We have mapped the magnetic field ($B$-field) for a region of about 30 pc around the centre of our Galaxy, which encompasses the circumnuclear disk (CND), the minispiral, and the 20 km s$^{-1}$ and 50 km s$^{-1}$ molecular clouds, using thermal dust polarization observations obtained from SOFIA/HAWC+ and JCMT/SCUPOL. We decompose the spectra of $^{12}$CO ($J\!=\!3\!\rightarrow\!2$) transition from this region into individual cloud components and find the polarization observed at different wavelengths might be tracing completely different layers of dust along the line of sight. We use modified Davis-Chandrasekhar-Fermi methods to measure the strength of $B$-field projected in the plane of the sky ($B_{{}_{\mathrm{POS}}}$). The mean $B_{{}_{\mathrm{POS}}}$ of the CND and the minispiral, probed at 53 $\mu$m is of the order of $\sim\!2$ mG. $B_{{}_{\mathrm{POS}}}\!\!\!<\!1$ mG close to the Galactic Centre, in the region of the ionized mini-cavity within the CND, and increases outwards. However, the longer wavelength polarization at 216 $\mu$m appears to come from a dust layer that is cooler and behind the CND and has a stronger $B$-field of about 7 mG. The $B$-field strength is lowest along the Eastern Arm of the minispiral, which is also the only region with Alfv\'en Mach number, $\mathcal{M}_{\mathrm{A}}>1$ and mass-to-flux ratio, $\lambda\!\gtrsim\!1$. Such an observed weak $B$-field could be a result of the low resolution of the observation, where the tangled $B$-fields due to the strong turbulence in the high density clumps of the CND are lost within the beam size of the observation., Comment: 19 pages, 18 figures, Accepted for Publication in MNRAS