Universidad de Los Andes
Repositorio institucional Séneca Universidad de los Andes (Colombia)Not a member yet
38709 research outputs found
Sort by
Detection and analysis of distance indicator stars using statistical learning
No full textModern astronomy faces fundamental challenges, including understanding cosmic evolution, refining distance measurement techniques, and processing vast astronomical datasets. A cornerstone in this effort is the cosmic distance ladder—a sequence of methods for measuring distances, where each rung builds upon the accuracy of the previous one. Among its most crucial components are Cepheid variable stars and carbon stars, which play a pivotal role in calibrating cosmic distances and mapping the Universe’s structure.
This dissertation examines statistical and computational approaches aimed at improving the reliability of the Period-Luminosity relation model for Cepheids and enhancing the detection of carbon stars in large-scale spectroscopic surveys. The central research question is: How can the Cepheids Period-Luminosity relation model be improved and the detection of carbon stars enhanced?
The first part of this study analyzes the impact of using median instead of mean magnitudes in determining the slope and zero-point of the Period-Luminosity relation. Using OGLE-IV survey data in the and bands, corrected for extinction, the research applies robust -regression, preferred over ordinary least squares due to influential data points violating statistical assumptions. The study identifies deficiencies in the Period-Luminosity relation model and successfully corrects them, revealing a clear break in the relation for Cepheids with periods longer than 10 days. Furthermore, it highlights biases introduced when using median magnitudes, especially in the band.
The second part focuses on developing a machine learning-based classification model to detect carbon stars in highly imbalanced and large datasets. Leveraging key spectral indices from LAMOST DR9 and color-magnitude data from Gaia DR3, a binary classification model was trained and validated using a modified Monte Carlo cross-validation approach developed in this study. The final model exhibits excellent generalization performance, achieving a validation correctness rate exceeding 97%.
After deployment, the model detected 6372 carbon stars across millions of samples in just hours, while accounting for feature uncertainties. A visual inspection of 563 candidates in the Orion region confirmed over 94% as genuine carbon stars, underscoring the effectiveness of the proposed methodology.
By refining Cepheid distance calibration and enhancing rare star detection through machine learning, this research contributes valuable improvements to cosmic distance
measurement techniques and large-scale stellar classification.DoctoradoStatistical learnin
Risk-Adjusted Social Discount Rate for Colombia
No full textLa tasa social de descuento es un parámetro para valorar y comparar los costos y beneficios económicos intertemporales de proyectos, programas o políticas públicas (PPP). En Colombia, la tasa oficial actual fue estimada por Piraquive et al. (2018) a partir de Harberger (1969a). Dicha estimación no incorpora de forma explícita el riesgo de los PPP, pues algunos presentan baja exposición al riesgo, mientras que otros muestran beneficios que se correlacionan con el ciclo económico, haciéndolos más riesgosos.
Este estudio propone un modelo para estimar una Tasa Social de Descuento Ajustada por Riesgo (TSDAR), según el grado de exposición de un PPP al ciclo económico. El modelo se calibra utilizando datos de cuentas nacionales y se contrasta con el trabajo de Harberger (1969a). Los resultados indican que la TSDAR aumenta con el riesgo y la aversión a éste, pero puede mostrar una trayectoria decreciente en escenarios de bajo crecimiento y alta aversión al riesgo. Para 2025, un proyecto cuyos beneficios evolucionan con el ciclo económico deben evaluarse con tasas que van desde 4.95 %–6.09 % en el año 1, a 5.85 %–7.07 % a el año 30, en lugar del 9 % actual. Estos resultados destacan la importancia de incluir la exposición cíclica de los PPP en la evaluación de proyectos en Colombia.The social discount rate is a parameter used to assess and compare the intertemporal economic costs and benefits of projects, programs, or public policies (PPP). In Colombia, the current official rate was estimated by Piraquive et al. (2018), based on Harberger (1969a). This estimation does not explicitly incorporate the risk of PPPs, despite the fact that some exhibit low risk exposure while others display benefits correlated with the economic cycle, making them riskier.
This study proposes a model to estimate a Risk-Adjusted Social Discount Rate (TSDAR), based on the extent to which a PPP is exposed to the economic cycle. The model is calibrated using national accounts data and is contrasted with the work of Harberger (1969a). The results indicate that the TSDAR increases with risk and risk aversion, but it may exhibit a declining trajectory over time under scenarios of low growth and high risk aversion. For 2025, projects with benefits correlated to the economic cycle should be evaluated using rates ranging from 4.95 %–6.09 % in year 1, to 5.85 %–7.07 % by year 30—significantly lower than the current official rate of 9 %. These results underscore the importance of incorporating the cyclical exposure of PPPs in the economic evaluation of projects Colombia.Maestrí
NLGN1 gene knockout modifies the behavioral and transcriptional response to stress in zebrafish larvae
No full textEarly-life stress exerts lasting effects on brain development and behavior, yet the underlying genetic modulators remain incompletely understood. In this study, we used CRISPR- Cas9 to nlgn1 knockout, a gene critical for excitatory synapse organization, to investigate its role in the behavioral and transcriptional response to chronic stress in zebrafish (Danio rerio) larvae. At 14 days post-fertilization, wild-type (WT) larvae exhibited a robust behavioral response to chronic unpredictable stress (CUS), characterized by increased locomotor activity and reduced immobility— hallmarks of an adaptive stress coping profile. In contrast, nlgn1 knockout (nlgn1 KO) larvae showed a blunted locomotor response and heightened immobility, suggesting impaired arousal and passive coping strategies under stress. At the transcriptomic level, WT larvae demonstrated coordinated upregulation of stress-responsive genes, including crhbp, fkbp5, klf9 (HPI axis), and bdnf, npas4a (synaptic plasticity), whereas nlgn1 KO larvae showed a reduced number of differentially expressed genes and attenuated activation of these pathways. Functional enrichment analysis confirmed that synaptic signaling and glucocorticoid-related routes were activated in WT but absent or muted in nlgn1 KO larvae. Together, our results reveal that nlgn1 is necessary for mounting both molecular and behavioral responses to early-life stress, positioning it as a key modulator of stress adaptation during neurodevelopment.MaestríaLaboratorio de Neurogenética y comportamient
Consultoría para Modak - To Grow Strategy
No full textModak has made financial education accessible for teenagers aged 13 to 17 but has identified a significant gap in reaching children under 13. This younger audience requires a different approach, tailored to their interests, cognitive abilities, and ways of learning. This design project introduces the To Grow strategy, a capsule made up of three key touchpoints (MiniMentor, What’s the Price?, and MoRecap) aimed at fostering exploration, reflection, and recognition in early financial learning.
Through this strategy, Modak is empowered to expand its impact by nurturing a positive relationship with money from an early age, built on emotional and everyday connections between children and their parents. To inform the project, interviews and surveys were conducted with both parents and children to better understand their needs, motivations, and expectations regarding financial learning. The findings revealed a crucial emotional bond between parents and children that can be leveraged to strengthen financial habits through real life experiences. Additionally, it was found that children are more motivated to learn when doing so in playful environments and alongside their caregivers.
As a result, an integrated strategy was developed to promote motivation, joint participation, and experiential financial learning in childhood. This project encourages Modak to pave the way for children to take risks, explore, make mistakes, learn, and build a healthy and confident relationship with money from a young age.Pregrad
A Hybrid Approach for the Container Loading Problem for Enhancing the Dynamic Stability Representation
No full textIn the container loading problem (CLP), the construction of packing patterns isdriven by the maximization of the volume occupied, and comprises several constraints suchas loading feasibility, weight balance, cargo stability, operational safety, material handling,and the prevention of cargo damage during container shipping. Previous works introduceddynamic stability indicators using simulation or statistical approaches. However, thisfirstly exponentially increases the computational burden, and secondly misrepresents theessential kinetic mechanical aspects. This paper presents a hybrid scheme to solve theCLP by embedding a mechanical model into a reactive GRASP algorithm, leading totwo main novelties; namely, the substitution of the physics simulation engine to find thedynamic stability of the packing patterns, and a modified structure of the metaheuristic,guaranteeing specified minimum stability while achieving efficient packing patterns. Themechanical model dynamically analyzes the forces and accelerations acting on the cargo topredict loss of support, overturning, or critical velocity deltas that would damage it. Atthe same time, the reactive GRASP algorithm considers the dynamic stability indicatorsin the improvement steps. The stability indicators are obtained from the mechanicalmodel, allowing the user to know the percentage of damaged boxes in a packing pattern.The effectiveness of the proposed approach is tested using a set of classical benchmarkinstances, obtaining adequately accurate solutions within a short computational time. Theresulting scheme integrates real-world problem conditions and achieves dynamic stabilitysolutions at an acceptable computational cost; it is programmed in C++ instead of relyingon proprietary simulation tools
Designing urban futures: a methodological framework for the spatial allocation of nature-based solutions through participatory planning and ecosystem services evaluation
No full textCities in the Global South, particularly in Latin America and the Caribbean are increasingly vulnerable to the compounded effects of climate change, unplanned urban expansion, and persistent socio-environmental inequalities. Rapid urbanization in these regions often occurs in contexts marked by institutional fragility, weak land governance, and limited technical or financial resources. As a result, Nature-based Solutions (NbS) have gained traction as cost-effective and multifunctional strategies for promoting climate resilience and sustainable urban development. However, the effective implementation of NbS remains constrained by key challenges: the lack of participatory planning frameworks, difficulties in selecting appropriate biophysical evaluation methods under data scarcity, and the limited use of economic valuation approaches to support informed decision-making.
This thesis addresses key knowledge and implementation gaps by developing and testing a methodological framework tailored to the realities of resource-constrained urban environments in the Global South. The framework comprises a feasibility assessment, a Decision Support Tool (DST), and the application of economic valuation methods. Grounded in empirical research conducted in Bogotá, Colombia—within the urban renewal project El Reencuentro—the thesis explores three core dimensions of NbS planning: (i) the integration of socio-economic perceptions and stakeholder participation to define spatial priorities and select appropriate NbS typologies; (ii) the development of a DST to guide the selection of biophysical assessment methods under local constraints (e.g., limited data, budget, and technical capacity); and (iii) the application and comparative analysis of three economic valuation approaches (data synthesis, behavior-based, and stated-preference methods) to assess NbS benefits in data-scarce contexts.
Each phase developed in this thesis follows a flexible, scalable, and context-sensitive structure, allowing their application at multiple urban scales (neighborhood, block, district) and under varying resource conditions. The methodology is based on a participatory framework for NbS location and selection, a biophysical evaluation tool for method selection, and a benefit economic valuation. The participatory framework incorporates citizens' preferences and socio-demographic factors to improve the social legitimacy and long-term relevance of NbS interventions. The biophysical evaluation tool introduces decision rules based on a systematic literature review of 256 studies, guiding the selection of feasible methods while avoiding overreliance on complex models in data-poor settings. The economic valuation analysis reveals the strengths and limitations of each method, underscoring the importance of climate awareness, education, and local knowledge in determining willingness to pay for ecosystem services. At a broader level, the thesis contributes to the growing call for inclusive, adaptable, and evidence-based planning approaches in the Global South. It demonstrates that simplified yet robust tools—grounded in participatory processes and aligned with local constraints—can enable cities to move beyond generic frameworks, transforming them into active agents of ecological restoration and social equity in the face of climate uncertainty.euPOLIS project: https://eupolis-project.eu/Doctorad
Neurobehavioral Effects of bdnf gene Silencing via Intracerebroventricular Delivery of Nanobioconjugated CRISPRi dCas9/KRAB in adult brain of female Poecilia reticulata
No full textThe development of effective gene modulation tools for use in adult organisms remains a critical challenge in neuroscience, particularly for studying gene function with temporal and spatial precision. In vivo gene silencing strategies, especially those avoiding permanent genetic modifications, are essential for understanding dynamic processes in the adult brain. In this context, the CRISPR interference (CRISPRi) system has emerged as a promising approach, enabling reversible and targeted gene repression. Among potential targets, brain-derived neurotrophic factor (bdnf) is a key regulator of neurodevelopment, synaptic plasticity, and behavior, with altered expression patterns linked to a variety of neurological and psychiatric disorders. This study aimed to evaluate the neurobehavioral effects of bdnf silencing in Poecilia reticulata (guppy) via intracerebroventricular (ICV) delivery of a nanobioconjugated CRISPRi-dCas9 system.
The CRISPRi system, consisting of a dCas9-KRAB complex and a specific sgRNA, was delivered as a linearized plasmid conjugated to magnetic nanoparticles (MNPs) and administered directly into the third ventricle using a ICV microinjection platform. Control groups (non-injected, saline-injected, and MNP injected) were analyzed at molecular levels to confirm that no changes in bdnf expression occurred in the absence of CRISPRi treatment, it showed no significant changes in bdnf expression (p = 0.924), ensuring that any observed effects were attributable solely to the gene silencing intervention. In contrast, fish treated with the CRISPRi system exhibited a modest downregulation of bdnf gene.
These findings validate the neutrality of the delivery system and support its application for temporally controlled gene modulation and contributes to the development of non-viral, reversible CRISPR-based tools to investigate gene function and behavioral outcomes in the adult brain.MaestríaGenética y edición genétic
Puerta del Saber Fucha
No full textEste proyecto, desarrollado durante el noveno semestre de formación académica, surge a partir de una pregunta fundamental: ¿Cuál es mi arquitectura?. A lo largo del semestre, esta reflexión se construyó desde tres conceptos clave: penumbra, matiz y contraste. Estos ejes conceptuales se transformaron en herramientas de exploración espacial, permitiendo descubrir los elementos esenciales que definen mi lenguaje arquitectónico.
El proceso inicia con la creación de un conjunto espacial primario -visible en las primeras imágenes - donde se materializa esta búsqueda personal. En esta etapa, se revelan elementos como la interacción entre luz cenital y tejidos laminares de madera, que generan atmósferas de transición entre sombra y claridad. Aparecen también las primeras decisiones materiales, donde el contraste entre madera y concreto comienza a definir la identidad del proyecto.
Con estos elementos conceptuales y materiales como base, surge la propuesta arquitectónica: la Biblioteca Puerta del Saber Fucha. Este espacio busca responder a la pregunta inicial, integrando los valores descubiertos en una arquitectura que es tanto íntima como expresiva, reflexiva y funcional.Pregrad
Automatic detection of multimodal humor in latin american spanish
No full textHumor in human communication is intrinsically multimodal, emerging from complex interactions among language, voice, facial expression, and affect. Yet computational studies have focused almost exclusively on English, leaving Spanish— and Latin-American varieties in particular—largely unexplored. We introduce LS-FUNNY, the first public corpus for multimodal humor detection in Latin-American Spanish, built from 272 TEDx talks and comprising 2040 balanced instances annotated via laughter markers. A unified pipeline extracts five complementary modalities: visual cues with OpenFace, acoustic descriptors with Librosa, contextual sentence embeddings from XLM-RoBERTa, affective scores from the NRC VAD lexicon, and a binary flag denoting figurative language detected by GPT-4o. We benchmark three off-the-shelf classifiers—SVM, XGBoost, and a feed-forward MLP—over all feature combinations. The best result (accuracy and F1 = 0.65) is obtained by XGBoost when fusing all modalities, confirming the benefit of multimodal integration. Remarkably, a zero-shot GPT-4o baseline fed with text only reaches 0.66 accuracy and 0.64 F1, highlighting both the strength of large language models and the residual cues missed without audio–visual context. We release the dataset and processing scripts to foster reproducible research and outline concrete paths for enlarging the corpus through automatic transcription and laughter detection.Pregrad
Optimization of Bus Dispatching in Public Transportation Through a Heuristic Approach Based on Passenger Demand Forecasting
No full textAccurate and adaptive bus dispatching is vital for medium-sized urban centers, where static schedules often fail to accommodate fluctuating passenger demand. In this work, we propose a dynamic heuristic that integrates machine learning-based demand forecasts into a discrete-time planning horizon, thereby enabling real-time adjustments to dispatch decisions. Additionally, we introduce a tailored mathematical model—grounded in mixed-integer linear programming and space-time flows—that serves as a benchmark to evaluate our heuristic’s performance under the operational constraints typical of traditional public transportation systems in Colombian mid-sized cities. A key contribution of this research lies in combining predictive modeling (using Prophet for passenger demand) with operational optimization, ensuring that dispatch frequencies adapt promptly to varying ridership levels. We validated our approach using a real-world case study in Montería (Colombia), covering eight representative routes over a full day (5:00–21:00). Numerical experiments show that: 1. Our heuristic matches or surpasses 95% of the optimal solution’s operational utility on most routes, with an average gap of 4.7%, relative to the benchmark mathematical model. 2. It maintains high service levels—above 90% demand coverage on demanding corridors—and robust bus utilization, without incurring excessive operating costs. 3. It reduces computation times by up to 98% compared to the optimization model, making it practically viable for daily scheduling where solving large-scale models exactly can be prohibitively time-consuming. Overall, these results underscore the heuristic’s practical effectiveness in boosting profitability, optimizing resource use, and rapidly adapting to demand fluctuations. The proposed framework thus serves as a scalable and implementable tool for transportation operators seeking data-driven dispatch solutions that balance operational efficiency and service quality