Fabio Augusto Gonzalez

In response to the challenges of the energy transition and the need to diversify Colombia’s electricity mix, this research project focused on studying the wind potential in high-mountain ecosystems in the Andean region, using meteorological data from IDEAM. The study looked at the possibility of generating small-scale electricity in isolated communities by applying the Weibull statistical distribution. A quantitative and exploratory approach was used. More than 70 data files were cleaned, filtered by percentiles, and validated. A Python-based tool was created to automate the data processing, distribution fitting, wind power estimation, and interactive visualization through a web application. In addition, data from NASA’s POWER API was included for comparison. The results showed that most stations had low wind potential (less than 50 W/m²), although some specific areas had better conditions and a better statistical fit using the NASA data.

Assessment of gastric distensibility is crucial for understanding various gastrointestinal disorders, particularly those affecting gastric motility and reflux conditions. Gastric distensibility can be altered by pathological conditions such as linitis plastica or gastric mucosal inflammation, as seen in Helicobacter pylori infections. Although distensibility is evaluated during routine endoscopy, the assessment remains purely visual. Moreover, some specialized endoscopic tools assess only specific gastric regions and are often unavailable. Artificial intelligence tools may used to estimate the gastric three-dimensional surface from the endoscopy video and capture temporal distensibility changes of the gastric surface. This paper introduces a novel spatio-temporal characterization of the gastric distensibility during a routine endoscopy procedure. The method reconstructs the gastric surface for each frame of a short endoscopy video using a shape-from- shading network trained with a synthetic endoscopy collection. Local and regional changes in surface bending are then quantified, which herein depend on the distensibility. This spatio-temporal characterization was applied to a group of 160 patients, among them 60 diagnosed with Helicobacter pylori confirmed by pathology examination. The obtained characteristics in three different regions: body, antrum, and pylorus, demonstrated significant differences between patients with and without Helicobacter pylori under a t-test (p < 0.05). The proposed method effectively captures subtle gastric wall bending from routine endoscopy videos, providing a strategy for analyzing distensibility in pathological conditions across various gastric regions without requiring specialized tools.

Mountainous areas face challenges such as rugged topography, harsh weather, and limited access to power grids; however, they also offer potential for renewable energy generation, mainly through solar and wind resources. This study aims to evaluate the feasibility of implementing renewable energy systems in these regions and identify the most studied renewable technologies in high mountain contexts using the PRISMA methodology for rigorous literature selection and VOSviewer for bibliometric analysis. Among them, solar photovoltaic and wind energy stand out due to their high potential in these environments. The study analyzes key parameters such as technological efficiency, solar radiation variability, and wind patterns, including technical aspects like minimum wind speeds derived from the Weibull distribution and solar irradiance levels necessary for system design. The results show that the insights obtained from the bibliometric analysis help evaluate the feasibility and performance of renewable energy solutions in complex terrains. In conclusion, the study highlights the most viable technologies for high mountain areas and provides recommendations for their implementation. Although technical and environmental challenges persist, these ecosystems offer significant opportunities for sustainable energy generation. The findings provide guidance for future research and the development of innovative projects in remote, mountainous regions.

Classification methods based on deep learning require selecting between fully-supervised or weakly-supervised approaches, each presenting limitations in uncertainty quantification and interpretability. A framework unifying both supervision modes while maintaining quantifiable interpretation metrics remains unexplored. We introduce WiSDoM (Weakly-Supervised Density Matrices), which uses kernel matrices to model probability distributions of input data and their labels. The framework integrates: (1) differentiable kernel density matrices enabling stochastic gradient descent optimization, (2) local-global attention mechanisms for multi-scale feature weighting, (3) data-driven prototype generation through kernel space sampling, and (4) ordinal regression through density matrix operations. WiSDoM was validated through supervised patch classification ( κ = 0.896) and weakly-supervised whole-slide classification ( κ = 0.930) on histopathology images. WiSDoM generates three quantifiable outputs: posterior probability distributions, variance-based uncertainty maps, and phenotype prototypes. Through validation in a Gleason grading task at a patch and whole-slide level using histopathology images, WiSDoM demonstrated consistent performance across supervision modes ( κ > 0.89) and prototype interpretability (0.88 expert agreement). These results show that kernel density matrices can serve as a foundation for classification models requiring both prediction interpretability and uncertainty quantification across supervision modes.

This paper introduces a novel approach to probabilistic deep learning, kernel density matrices, which provide a simpler yet effective mechanism for representing joint probability distributions of both continuous and discrete random variables. In quantum mechanics, a density matrix is the most general way to describe the state of a quantum system. This work extends the concept of density matrices by allowing them to be defined in a reproducing kernel Hilbert space. This abstraction allows the construction of differentiable models for density estimation, inference, and sampling, and enables their integration into end-to-end deep neural models. In doing so, we provide a versatile representation of marginal and joint probability distributions that allows us to develop a differentiable, compositional, and reversible inference procedure that covers a wide range of machine learning tasks, including density estimation, discriminative learning, and generative modeling. The broad applicability of the framework is illustrated by two examples: an image classification model that can be naturally transformed into a conditional generative model, and a model for learning with label proportions that demonstrates the framework’s ability to deal with uncertainty in the training samples. The framework is implemented as a library and is available at: https://github.com/fagonzalezo/kdm .

Anomaly detection in data streams consists of detecting abnormal data samples in continuous data streams. This poses challenges that traditional and advanced anomaly detection methods are not equipped to deal with, such as the occurrence of conceptual drift and the need for continuous learning. Current state-of-the-art methods for detecting anomalies in data streams rely on fixed memory using hash functions or nearest neighbors. However, these methods can struggle to handle high-frequency data points or effectively identify seamless outliers, and also, they cannot be trained within a comprehensive deep-learning framework. In our research, we propose a novel method for streaming anomaly detection based on incremental density estimation. The method uses random Fourier features and incorporates the principles of quantum measurements and density matrices. We present the definition of the method and a variant of it, which can be represented as an exponential moving average density and a simple moving average, respectively. The method is designed to handle potentially infinite data streams and has a constant update complexity of O (1). To evaluate its effectiveness, we performed a comprehensive analysis by comparing the two variants with 12 state-of-the-art algorithms for detecting anomalies in data streams. We use 12 different streaming datasets to ensure an extensive evaluation of performance and efficiency.