Search

To test the hypothesis that megaparsec-scale giant radio galaxies (GRGs) have multiple epochs of recurrent activity which lead to their giant sizes and to understand the nature of double-double radio galaxies (DDRGs), we have built the largest sample of giant DDRGs from the LOFAR Two Metre Sky Survey (LoTSS). The sample consists of 111 sources, of which 76 are newly identified. Their redshift and total projected size ranges are between 0.06 to 1.6 and 0.7 Mpc to 3.2 Mpc, respectively. We conducted a detailed analysis to characterise their properties, including arm-length ratios, flux density ratios of pairs of lobes, and misalignment angles. These measurements allow us to study the symmetry parameters, which are influenced by the immediate and large-scale environments of DDRGs. Our study conclusively shows that the cocoons in which the inner lobes of DDRGs grow are often contaminated with surrounding material from the external medium, leading to observed asymmetries in the inner lobes. Our analysis also reveals highly misaligned DDRGs, which could be due to environmental factors and/or changes in the supermassive black hole jet ejection axes. By studying the misalignment angles, we assess the stability of the jets in these systems in relation to their environment. Additionally, we have also characterised the large-scale environments of DDRGs, identifying them within dense galaxy clusters and studying how cluster weather significantly affects their morphologies. Interestingly, we have identified two gigahertz peaked-spectrum (GPS) candidates in the unresolved cores of the DDRGs, as well as one triple-double candidate, which, if confirmed, would be only the fifth known case. Additionally, we have discovered a DDRG in a distant galaxy cluster at $z\sim$\,1.4. Abridged., Comment: Submitted to A&A journal. 24 Pages. Comments are welcome

Let $\sigma : \mathbb C^d \rightarrow \mathbb C^d$ be an affine-linear involution such that $J_\sigma = -1$ and let $U, V$ be two domains in $\mathbb C^d.$ Let $\phi : U \rightarrow V$ be a $\sigma$-invariant $2$-proper map such that $J_\phi$ is affine-linear and let $\mathscr H(U)$ be a $\sigma$-invariant reproducing kernel Hilbert space of complex-valued holomorphic functions on $U.$ It is shown that the space $\mathscr H_\phi(V):=\{f \in \mathrm{Hol}(V) : J_\phi \cdot f \circ \phi \in \mathscr H(U)\}$ endowed with the norm $\|f\|_\phi :=\|J_\phi \cdot f \circ \phi\|_{\mathscr H(U)}$ is a reproducing kernel Hilbert space and the linear mapping $\varGamma_\phi$ defined by $ \varGamma_\phi(f) = J_\phi \cdot f \circ \phi,$ $f \in \mathrm{Hol}(V),$ is a unitary from $\mathscr H_\phi(V)$ onto $\{f \in \mathscr H(U) : f = -f \circ \sigma\}.$ Moreover, a neat formula for the reproducing kernel $\kappa_{\phi}$ of $\mathscr H_\phi(V)$ in terms of the reproducing kernel of $\mathscr H(U)$ is given. The above scheme is applicable to symmetrized bidisc, tetrablock, $d$-dimensional fat Hartogs triangle and $d$-dimensional egg domain. Although some of these are known, this allows us to obtain an analog of von Neumann's inequality for contractive tuples naturally associated with these domains., Comment: 14 pages and 1 table

We introduce a family of Helson matrices induced by the Laplace transform of a class of regular positive Borel measures $\mu,$ not necessarily finite, on $(0, \infty)$ and discuss their boundedness, Schatten-class properties and scattering theory. We also briefly discuss the class of Helson matrices induced by signed measures., Comment: 20 pages

We present the results of a nation-wide baseline survey, conducted by us, for the status of Astronomy education among secondary school students in India. The survey was administered in 10 different languages to over 2000 students from diverse backgrounds, and it explored multiple facets of their perspectives on astronomy. The topics included students' views on the incorporation of astronomy in curricula, their grasp of fundamental astronomical concepts, access to educational resources, cultural connections to astronomy, and their levels of interest and aspirations in the subject. We find notable deficiencies in students' knowledge of basic astronomical principles, with only a minority demonstrating proficiency in key areas such as celestial sizes, distances, and lunar phases. Furthermore, access to resources such as telescopes and planetariums remain limited across the country. Despite these challenges, a significant majority of students expressed a keen interest in astronomy. We further analyze the data along socioeconomic and gender lines. Particularly striking were the socioeconomic disparities, with students from resource-poor backgrounds often having lower levels of access and proficiency. Some differences were observed between genders, although not very pronounced. The insights gleaned from this study hold valuable implications for the development of a more robust astronomy curriculum and the design of effective teacher training programs in the future., Comment: 15 pages, 19 figures

Stopwords are commonly used words in a language that are often considered to be of little value in determining the meaning or significance of a document. These words occur frequently in most texts and don't provide much useful information for tasks like sentiment analysis and text classification. English, which is a high-resource language, takes advantage of the availability of stopwords, whereas low-resource Indian languages like Marathi are very limited, standardized, and can be used in available packages, but the number of available words in those packages is low. Our work targets the curation of stopwords in the Marathi language using the MahaCorpus, with 24.8 million sentences. We make use of the TF-IDF approach coupled with human evaluation to curate a strong stopword list of 400 words. We apply the stop word removal to the text classification task and show its efficacy. The work also presents a simple recipe for stopword curation in a low-resource language. The stopwords are integrated into the mahaNLP library and publicly available on https://github.com/l3cube-pune/MarathiNLP ., Comment: Accepted at I2CT 2024

Egocentric vision aims to capture and analyse the world from the first-person perspective. We explore the possibilities for egocentric wearable devices to improve and enhance industrial use cases w.r.t. data collection, annotation, labelling and downstream applications. This would contribute to easier data collection and allow users to provide additional context. We envision that this approach could serve as a supplement to the traditional industrial Machine Vision workflow. Code, Dataset and related resources will be available at: https://github.com/Vivek9Chavan/EgoVis24, Comment: To be presented at the First Joint Egocentric Vision (EgoVis) Workshop, held in conjunction with CVPR 2024

Low-rank approximations, of the weight and feature space can enhance the performance of deep learning models, whether in terms of improving generalization or reducing the latency of inference. However, there is no clear consensus yet on \emph{how}, \emph{when} and \emph{why} these approximations are helpful for large language models (LLMs). In this work, we empirically study the efficacy of weight and feature space decomposition in transformer-based LLMs. We demonstrate that surgical decomposition not only provides critical insights into the trade-off between compression and language modelling performance, but also sometimes enhances commonsense reasoning performance of LLMs. Our empirical analysis identifies specific network segments that intrinsically exhibit a low-rank structure. Furthermore, we extend our investigation to the implications of low-rank approximations on model bias. Overall, our findings offer a novel perspective on optimizing LLMs, presenting the low-rank approximation not only as a tool for performance enhancements, but also as a means to potentially rectify biases within these models. Our code is available at \href{https://github.com/nyunAI/SFSD-LLM}{GitHub}., Comment: Accepted at ACL 2024

The growing popularity of machine learning (ML) and the integration of ML components with other software artifacts has led to the use of continuous integration and delivery (CI/CD) tools, such as Travis CI, GitHub Actions, etc. that enable faster integration and testing for ML projects. Such CI/CD configurations and services require synchronization during the life cycle of the projects. Several works discussed how CI/CD configuration and services change during their usage in traditional software systems. However, there is very limited knowledge of how CI/CD configuration and services change in ML projects. To fill this knowledge gap, this work presents the first empirical analysis of how CI/CD configuration evolves for ML software systems. We manually analyzed 343 commits collected from 508 open-source ML projects to identify common CI/CD configuration change categories in ML projects and devised a taxonomy of 14 co-changes in CI/CD and ML components. Moreover, we developed a CI/CD configuration change clustering tool that identified frequent CI/CD configuration change patterns in 15,634 commits. Furthermore, we measured the expertise of ML developers who modify CI/CD configurations. Based on this analysis, we found that 61.8% of commits include a change to the build policy and minimal changes related to performance and maintainability compared to general open-source projects. Additionally, the co-evolution analysis identified that CI/CD configurations, in many cases, changed unnecessarily due to bad practices such as the direct inclusion of dependencies and a lack of usage of standardized testing frameworks. More practices were found through the change patterns analysis consisting of using deprecated settings and reliance on a generic build language. Finally, our developer's expertise analysis suggests that experienced developers are more inclined to modify CI/CD configurations.

We show the (anti)holomorphicity of smooth Kobayashi isometries of the diamond, the domain defined as $\triangle:=\{z\in\mathbb C^2:|z_1|+|z_2|<1\}$. Additionally, we discuss the problem of uniqueness of real geodesics, left inverses and strict convexity of indicatrices., Comment: 15 pages

Conventional scaling of neural networks typically involves designing a base network and growing different dimensions like width, depth, etc. of the same by some predefined scaling factors. We introduce an automated scaling approach leveraging second-order loss landscape information. Our method is flexible towards skip connections a mainstay in modern vision transformers. Our training-aware method jointly scales and trains transformers without additional training iterations. Motivated by the hypothesis that not all neurons need uniform depth complexity, our approach embraces depth heterogeneity. Extensive evaluations on DeiT-S with ImageNet100 show a 2.5% accuracy gain and 10% parameter efficiency improvement over conventional scaling. Scaled networks demonstrate superior performance upon training small scale datasets from scratch. We introduce the first intact scaling mechanism for vision transformers, a step towards efficient model scaling., Comment: Accepted At ICLR 2024 (Tiny Paper Track)

Dynamic 2-[18F] fluoro-2-deoxy-D-glucose positron emission tomography (dFDG-PET) for human brain imaging has considerable clinical potential, yet its utilization remains limited. A key challenge in the quantitative analysis of dFDG-PET is characterizing a patient-specific blood input function, traditionally reliant on invasive arterial blood sampling. This research introduces a novel approach employing non-invasive deep learning model-based computations from the internal carotid arteries (ICA) with partial volume (PV) corrections, thereby eliminating the need for invasive arterial sampling. We present an end-to-end pipeline incorporating a 3D U-Net based ICA-net for ICA segmentation, alongside a Recurrent Neural Network (RNN) based MCIF-net for the derivation of a model-corrected blood input function (MCIF) with PV corrections. The developed 3D U-Net and RNN was trained and validated using a 5-fold cross-validation approach on 50 human brain FDG PET datasets. The ICA-net achieved an average Dice score of 82.18% and an Intersection over Union of 68.54% across all tested scans. Furthermore, the MCIF-net exhibited a minimal root mean squared error of 0.0052. The application of this pipeline to ground truth data for dFDG-PET brain scans resulted in the precise localization of seizure onset regions, which contributed to a successful clinical outcome, with the patient achieving a seizure-free state after treatment. These results underscore the efficacy of the ICA-net and MCIF-net deep learning pipeline in learning the ICA structure's distribution and automating MCIF computation with PV corrections. This advancement marks a significant leap in non-invasive neuroimaging.

Despite the impressive performance of LLMs, their widespread adoption faces challenges due to substantial computational and memory requirements during inference. Recent advancements in model compression and system-level optimization methods aim to enhance LLM inference. This survey offers an overview of these methods, emphasizing recent developments. Through experiments on LLaMA(/2)-7B, we evaluate various compression techniques, providing practical insights for efficient LLM deployment in a unified setting. The empirical analysis on LLaMA(/2)-7B highlights the effectiveness of these methods. Drawing from survey insights, we identify current limitations and discuss potential future directions to improve LLM inference efficiency. We release the codebase to reproduce the results presented in this paper at https://github.com/nyunAI/Faster-LLM-Survey, Comment: Accepted at IJCAI '24 (Survey Track), Updated TGI results

Large Language Models (LLMs) are adept at text manipulation -- tasks such as machine translation and text summarization. However, these models can also be prone to hallucination, which can be detrimental to the faithfulness of any answers that the model provides. Recent works in combating hallucinations in LLMs deal with identifying hallucinated sentences and categorizing the different ways in which models hallucinate. This paper takes a deep dive into LLM behavior with respect to hallucinations, defines a token-level approach to identifying different kinds of hallucinations, and further utilizes this token-level tagging to improve the interpretability and faithfulness of LLMs in dialogue summarization tasks. Through this, the paper presents a new, enhanced dataset and a new training paradigm., Comment: All authors contributed equally to this work

Recently introduced ControlNet has the ability to steer the text-driven image generation process with geometric input such as human 2D pose, or edge features. While ControlNet provides control over the geometric form of the instances in the generated image, it lacks the capability to dictate the visual appearance of each instance. We present FineControlNet to provide fine control over each instance's appearance while maintaining the precise pose control capability. Specifically, we develop and demonstrate FineControlNet with geometric control via human pose images and appearance control via instance-level text prompts. The spatial alignment of instance-specific text prompts and 2D poses in latent space enables the fine control capabilities of FineControlNet. We evaluate the performance of FineControlNet with rigorous comparison against state-of-the-art pose-conditioned text-to-image diffusion models. FineControlNet achieves superior performance in generating images that follow the user-provided instance-specific text prompts and poses compared with existing methods. Project webpage: https://samsunglabs.github.io/FineControlNet-project-page, Comment: Hongsuk Choi and Isaac Kasahara have eqaul contributions. 19 pages, 15 figures, 3 tables

Due to the substantial scale of Large Language Models (LLMs), the direct application of conventional compression methodologies proves impractical. The computational demands associated with even minimal gradient updates present challenges, particularly on consumer-grade hardware. This paper introduces an innovative approach for the parametric and practical compression of LLMs based on reduced order modelling, which entails low-rank decomposition within the feature space and re-parameterization in the weight space. Notably, this compression technique operates in a layer-wise manner, obviating the need for a GPU device and enabling the compression of billion-scale models within stringent constraints of both memory and time. Our method represents a significant advancement in model compression by leveraging matrix decomposition, demonstrating superior efficacy compared to the prevailing state-of-the-art structured pruning method., Comment: Brief technical report; Code will be made available at https://github.com/transmuteAI/trailmet/tree/main/trailmet/algorithms/llm-rom

Intimacy estimation of a given text has recently gained importance due to the increase in direct interaction of NLP systems with humans. Intimacy is an important aspect of natural language and has a substantial impact on our everyday communication. Thus the level of intimacy can provide us with deeper insights and richer semantics of conversations. In this paper, we present our work on the SemEval shared task 9 on predicting the level of intimacy for the given text. The dataset consists of tweets in ten languages, out of which only six are available in the training dataset. We conduct several experiments and show that an ensemble of multilingual models along with a language-specific monolingual model has the best performance. We also evaluate other data augmentation methods such as translation and present the results. Lastly, we study the results thoroughly and present some noteworthy insights into this problem.

This paper presents our approach for the WASSA 2023 Empathy, Emotion and Personality Shared Task. Empathy and distress are human feelings that are implicitly expressed in natural discourses. Empathy and distress detection are crucial challenges in Natural Language Processing that can aid our understanding of conversations. The provided dataset consists of several long-text examples in the English language, with each example associated with a numeric score for empathy and distress. We experiment with several BERT-based models as a part of our approach. We also try various ensemble methods. Our final submission has a Pearson's r score of 0.346, placing us third in the empathy and distress detection subtask., Comment: Accepted at the WASSA 2023 workshop at ACL 2023

Evaporation Residue (ER) cross-sections and ER-gated $\gamma$-ray fold distributions are measured for the $^{32}$S + $^{154}$Sm nuclear reaction above the Coulomb barrier at six different beam energies from 148 to 191 MeV. $\gamma$-ray multiplicities and spin distributions are extracted from the ER-gated fold distributions. The ER cross-sections measured in the present work are found to be much higher than what was reported in a previous work using a very different target-projectile ($^{48}$Ti + $^{138}$Ba) combination, leading to the same compound nucleus $^{186}$Pt, with much less mass asymmetry in the entrance channel than the present reaction. This clearly demonstrates the effect of the entrance channel on ER production cross-section. The ER cross-sections measured in the present work are compared with the results of both the statistical model calculations and the dynamical model calculations. Statistical model calculations have been performed to generate a range of parameter space for both the barrier height and Kramers' viscosity parameter over which the ER cross-section data can be reproduced. The calculations performed using the dinuclear system (DNS) model reproduce the data considering both complete and incomplete fusion processes. DNS calculations indicate the need for the inclusion of incomplete fusion channel at higher energies to reproduce the ER cross-sections., Comment: 13 pages, 18 figures

We study the problem of aligning a video that captures a local portion of an environment to the 2D LiDAR scan of the entire environment. We introduce a method (VioLA) that starts with building a semantic map of the local scene from the image sequence, then extracts points at a fixed height for registering to the LiDAR map. Due to reconstruction errors or partial coverage of the camera scan, the reconstructed semantic map may not contain sufficient information for registration. To address this problem, VioLA makes use of a pre-trained text-to-image inpainting model paired with a depth completion model for filling in the missing scene content in a geometrically consistent fashion to support pose registration. We evaluate VioLA on two real-world RGB-D benchmarks, as well as a self-captured dataset of a large office scene. Notably, our proposed scene completion module improves the pose registration performance by up to 20%., Comment: 8 pages