Search

In this study, we develop and assess new corpus selection and training methodologies to improve the effectiveness of Turkish language models. Specifically, we adapted Large Language Model generated datasets and translated English datasets into Turkish, integrating these resources into the training process. This approach led to substantial enhancements in model accuracy for both few-shot and zero-shot learning scenarios. Furthermore, the merging of these adapted models was found to markedly improve their performance. Human evaluative metrics, including task-specific performance assessments, further demonstrated that these adapted models possess a greater aptitude for comprehending the Turkish language and addressing logic-based queries. This research underscores the importance of refining corpus selection strategies to optimize the performance of multilingual models, particularly for under-resourced languages like Turkish., Comment: 2024 Innovations in Intelligent Systems and Applications Conference (ASYU)

In this study, a Turkish visual instruction model was developed and various model architectures and dataset combinations were analysed to improve the performance of this model. The Cosmos-LLaVA model, which is built by combining different large language models and image coders, is designed to overcome the deficiencies in the Turkish language. In the experiments, the effects of fine-tuning with various datasets on the model performance are analysed in detail. The results show that model architecture and dataset selection have a significant impact on performance. Bu \c{c}al{\i}\c{s}mada bir T\"urk\c{c}e g\"orsel talimat modeli geli\c{s}tirilerek bu modelin performans{\i}n{\i} art{\i}rmaya y\"onelik \c{c}e\c{s}itli model mimarileri ve veri k\"umesi kombinasyonlar{\i} derinlemesine incelenmi\c{s}tir. Farkl{\i} b\"uy\"uk dil modelleri ve g\"or\"unt\"u kodlay{\i}c{\i}lar{\i}n{\i}n bir araya getirilmesiyle olu\c{s}turulan Cosmos-LLaVA modeli, T\"urk\c{c}e dilindeki eksiklikleri gidermeye y\"onelik olarak tasarlanm{\i}\c{s}t{\i}r. Yap{\i}lan deneylerde, \c{c}e\c{s}itli veri k\"umeleri ile yap{\i}lan ince ayarlar{\i}n model performans{\i}n{\i} nas{\i}l etkiledi\u{g}i detayl{\i} olarak ele al{\i}nm{\i}\c{s}t{\i}r. Sonu\c{c}lar, model mimarisi ve veri k\"umesi se\c{c}iminin performans \"uzerinde \"onemli bir etkiye sahip oldu\u{g}unu g\"ostermektedir., Comment: in Turkish language, 2024 8th International Artificial Intelligence and Data Processing Symposium (IDAP)

While teleparallel $f(T)$ gravity has shown considerable potential in addressing cosmological tensions, such as the $H_0$ and $S_8$ discrepancies, we explore previously overlooked solution spaces within this framework that hold further promise. Specifically, we examine the case where the customary assumption of a strictly positive effective dark energy (DE) density -- natural in general relativity -- may not apply, offering new possibilities. Focusing on the exponential infrared model $f(T)=Te^{\beta T_0/T}$, where $T$ is the torsion scalar, we investigate cosmological solutions characterized by the constant $\beta$, which is controlled by the present-day matter density parameter $\Omega_{\rm m0}$. Ensuring consistency with CMB data, we find that the widely studied $\beta_{+}$ (positive exponent) case exhibits phantom behavior, while the previously overlooked $\beta_{-}$ (negative exponent) case -- sufficient to avoid instabilities or ghosts -- features a sign-changing DE density that transitions smoothly from negative to positive values at redshift $z_{\dagger} \sim 1.5$, consistent with recent approaches to alleviating cosmological tensions. Though the sign-changing DE density in the $f(T)$ model leads to a larger-than-expected enhancement, we further extend the analysis by incorporating a cosmological constant, $\Lambda$. This extension, $f(T) \rightarrow f(T) + 2\Lambda$, broadens the solution space consistent with the SH0ES $H_0$ measurement while maintaining consistency with CMB power spectra. Additionally, it introduces richer phenomenological possibilities, including the potential moderation or cessation of cosmic acceleration at very low redshifts, aligning with recent observational analyses, such as those from DESI BAO data. Our findings also suggest that existing $f(T)$ cosmological models should be revisited in light of the novel theoretical insights presented here., Comment: 22 pages, 8 figures, and no tables

Unraveling the complexities of Gene Regulatory Networks (GRNs) is crucial for understanding cellular processes and disease mechanisms. Traditional computational methods often struggle with the dynamic nature of these networks. This study explores the use of Graph Neural Networks (GNNs), a powerful approach for modeling graph-structured data like GRNs. Utilizing a Graph Attention Network v2 (GATv2), our study presents a novel approach to the construction and interrogation of GRNs, informed by gene expression data and Boolean models derived from literature. The model's adeptness in accurately predicting regulatory interactions and pinpointing key regulators is attributed to advanced attention mechanisms, a hallmark of the GNN framework. These insights suggest that GNNs are primed to revolutionize GRN analysis, addressing traditional limitations and offering richer biological insights. The success of GNNs, as highlighted by our model's reliance on high-quality data, calls for enhanced data collection methods to sustain progress. The integration of GNNs in GRN research is set to pioneer developments in personalized medicine, drug discovery, and our grasp of biological systems, bolstered by the structural analysis of networks for improved node and edge prediction., Comment: 24 Pages, 6 Figures

Infinitesimal light bundles on curved spacetimes can be studied via a Hamiltonian formalism, similar to the Newtonian paraxial rays. In this work, we assign a classical wave function to a thin null bundle and study its evolution equation. This is achieved via the usage of the Schr\"odinger operators within a procedure analogous to the one in the semi-classical regime of quantum mechanics. The correspondence between the metaplectic operators and the symplectic phase space transformations of the geodesic deviation variables is at the core of our method. It allows for the introduction of unitary operators. We provide two solutions of the null bundle wave function which differ by their origin: (i) a point source, and (ii) a finite source. It is shown that while the former wave function includes the same information as the standard thin null bundle framework, the latter is a Gaussian beam. The Gaussianity of the intensity profile of our beam depends on the spacetime curvature and not on the random processes. We show that this beam avoids the caustics of an instantaneous wavefront. Our results are applicable for any spacetime and they can be used to model light propagation from coherent sources while averting the mathematical singularities of the standard thin null bundle formalism. This is especially relevant when estimating cosmological distances in a realistic inhomogeneous universe., Comment: 31 pages, 4 figures

Regular patterns of vegetation are considered widespread landscapes, although their global extent has never been estimated. Among them, spotted landscapes are of particular interest in the context of climate change. Indeed, regularly spaced vegetation spots in semi-arid shrublands result from extreme resource depletion and prefigure catastrophic shift of the ecosystem to a homogeneous desert, while termite mounds also producing spotted landscapes were shown to increase robustness to climate change. Yet, their identification at large scale calls for automatic methods, for instance using the popular deep learning framework, able to cope with a vast amount of remote sensing data, e.g., optical satellite imagery. In this paper, we tackle this problem and benchmark some state-of-the-art deep networks on several landscapes and geographical areas. Despite the promising results we obtained, we found that more research is needed to be able to map automatically these earth mounds from space., Comment: 6 pages, 4 figures, 3 tables

In this study, we tackle the challenge of identifying plant species from ultra high resolution (UHR) remote sensing images. Our approach involves introducing an RGB remote sensing dataset, characterized by millimeter-level spatial resolution, meticulously curated through several field expeditions across a mountainous region in France covering various landscapes. The task of plant species identification is framed as a semantic segmentation problem for its practical and efficient implementation across vast geographical areas. However, when dealing with segmentation masks, we confront instances where distinguishing boundaries between plant species and their background is challenging. We tackle this issue by introducing a fuzzy loss within the segmentation model. Instead of utilizing one-hot encoded ground truth (GT), our model incorporates Gaussian filter refined GT, introducing stochasticity during training. First experimental results obtained on both our UHR dataset and a public dataset are presented, showing the relevance of the proposed methodology, as well as the need for future improvement., Comment: 5 pages, 5 figures, 2 tables

In recent years, mathematical models have become an indispensable tool in the planning, evaluation, and implementation of public health interventions. Models must often provide detailed information for many levels of population stratification. Such detail comes at a price: in addition to the computational costs, the number of considered input parameters can be large, making effective study design difficult. To address these difficulties, we propose a novel technique to reduce the dimension of the model input space to simplify model-informed intervention planning. The method works by first applying a dimension reduction technique on the model output space. We then develop a method which allows us to map each reduced output to a corresponding vector in the input space, thereby reducing its dimension. We apply the method to the HIV Optimization and Prevention Economics (HOPE) model, to validate the approach and establish proof of concept.

This paper introduces a generalized mean-based C^1-smooth robustness measure over discrete-time signals (D-GMSR) for signal temporal logic (STL) specifications. In conjunction with its C1-smoothness, D-GMSR is proven to be both sound and complete. Furthermore, it demonstrates favorable gradient properties and addresses locality and masking problems, which are critical for numerical optimization. The C^1-smoothness of the proposed formulations enables the implementation of robust and efficient numerical optimization algorithms to solve problems with STL specifications while preserving their theoretical guarantees. The practical utility of the proposed robustness measure is demonstrated on two real-world trajectory optimization problems: i) quadrotor flight, and ii) autonomous rocket landing. A sequential convex programming (SCP) framework, incorporating a convergence-guaranteed optimization algorithm (the prox-linear method) is used to solve inherently non-convex trajectory optimization problems with STL specifications. The implementation is available at https://github.com/UW-ACL/D-GMSR

Atmospheric turbulence poses a significant challenge to the performance of object detection models. Turbulence causes distortions, blurring, and noise in images by bending and scattering light rays due to variations in the refractive index of air. This results in non-rigid geometric distortions and temporal fluctuations in the electromagnetic radiation received by optical systems. This paper explores the effectiveness of turbulence image augmentation techniques in improving the accuracy and robustness of thermal-adapted and deep learning-based object detection models under atmospheric turbulence. Three distinct approximation-based turbulence simulators (geometric, Zernike-based, and P2S) are employed to generate turbulent training and test datasets. The performance of three state-of-the-art deep learning-based object detection models: RTMDet-x, DINO-4scale, and YOLOv8-x, is employed on these turbulent datasets with and without turbulence augmentation during training. The results demonstrate that utilizing turbulence-specific augmentations during model training can significantly improve detection accuracy and robustness against distorted turbulent images. Turbulence augmentation enhances performance even for a non-turbulent test set.

In this paper we prove a number of exact relations between optical observables, such as trigonometric parallax, position drift and the proper motion of a luminous source in addition to the variations of redshift and the viewing angle. These relations are valid in general relativity for any spacetime and they are of potential interest for astrometry and precise cosmology. They generalize the well-known Etherington's reciprocity relation between the angular diameter distance and the luminosity distance. Similar to the Etherington's relation, they hold independently of the spacetime metric, the positions and the motions of a light source or an observer. We show that those relations follow from the symplectic property of the bi-local geodesic operator, i.e., the geometric object that describes the light propagation between two distant regions of a spacetime. The set of relations we present is complete in the sense that no other relations between those observables should hold in general. In the meantime, we develop the mathematical machinery of the bi-local approach to light propagation in general relativity and its corresponding Hamiltonian formalism., Comment: 27 pages, 4 figures, 2 tables

We propose a nonlinear model predictive control (NMPC) framework based on a direct optimal control method that ensures continuous-time constraint satisfaction and accurate evaluation of the running cost, without compromising computational efficiency. We leverage the recently proposed successive convexification framework for trajectory optimization, where: (1) the path constraints and running cost are equivalently reformulated by augmenting the system dynamics, (2) multiple shooting is used for exact discretization, and (3) a convergence-guaranteed sequential convex programming (SCP) algorithm, the prox-linear method, is used to solve the discretized receding-horizon optimal control problems. The resulting NMPC framework is computationally efficient, owing to its support for warm-starting and premature termination of SCP, and its reliance on first-order information only. We demonstrate the effectiveness of the proposed NMPC framework by means of a numerical example with reference-tracking and obstacle avoidance. The implementation is available at https://github.com/UW-ACL/nmpc-ctcs

The number of open source language models that can produce Turkish is increasing day by day, as in other languages. In order to create the basic versions of such models, the training of multilingual models is usually continued with Turkish corpora. The alternative is to train the model with only Turkish corpora. In this study, we first introduce the cosmosGPT models that we created with this alternative method. Then, we introduce new finetune datasets for basic language models to fulfill user requests and new evaluation datasets for measuring the capabilities of Turkish language models. Finally, a comprehensive comparison of the adapted Turkish language models on different capabilities is presented. The results show that the language models we built with the monolingual corpus have promising performance despite being about 10 times smaller than the others.

The developments that language models have provided in fulfilling almost all kinds of tasks have attracted the attention of not only researchers but also the society and have enabled them to become products. There are commercially successful language models available. However, users may prefer open-source language models due to cost, data privacy, or regulations. Yet, despite the increasing number of these models, there is no comprehensive comparison of their performance for Turkish. This study aims to fill this gap in the literature. A comparison is made among seven selected language models based on their contextual learning and question-answering abilities. Turkish datasets for contextual learning and question-answering were prepared, and both automatic and human evaluations were conducted. The results show that for question-answering, continuing pretraining before fine-tuning with instructional datasets is more successful in adapting multilingual models to Turkish and that in-context learning performances do not much related to question-answering performances., Comment: in Turkish language. Baz{\i} \c{c}al{\i}\c{s}malar{\i} i\c{c}ermedi\u{g}ini s\"oyleyen hakem yorumu nedeniyle bir konferanstan kabul almad{\i}. Ancak hakemin bahsetti\u{g}i \c{c}al{\i}\c{s}malar bildiri g\"onderme son tarihinde yay{\i}nlanmam{\i}\c{s}t{\i}

We present successive convexification, a real-time-capable solution method for nonconvex trajectory optimization, with continuous-time constraint satisfaction and guaranteed convergence, that only requires first-order information. The proposed framework combines several key methods to solve a large class of nonlinear optimal control problems: (i) exterior penalty-based reformulation of the path constraints; (ii) generalized time-dilation; (iii) multiple-shooting discretization; (iv) $\ell_1$ exact penalization of the nonconvex constraints; and (v) the prox-linear method, a sequential convex programming (SCP) algorithm for convex-composite minimization. The reformulation of the path constraints enables continuous-time constraint satisfaction even on sparse discretization grids and obviates the need for mesh refinement heuristics. Through the prox-linear method, we guarantee convergence of the solution method to stationary points of the penalized problem and guarantee that the converged solutions that are feasible with respect to the discretized and control-parameterized optimal control problem are also Karush-Kuhn-Tucker (KKT) points. Furthermore, we highlight the specialization of this property to global minimizers of convex optimal control problems, wherein the reformulated path constraints cannot be represented by canonical cones, i.e., in the form required by existing convex optimization solvers. In addition to theoretical analysis, we demonstrate the effectiveness and real-time capability of the proposed framework with numerical examples based on popular optimal control applications: dynamic obstacle avoidance and rocket landing.