Los agentes de inferencia causal son sistemas inteligentes diseñados para descubrir y razonar sobre relaciones de causa y efecto mediante la interacción activa con su entorno. A diferencia de los modelos puramente predictivos, estos agentes no se limitan a detectar relaciones estadísticas en datos observacionales, sino que ejecutan intervenciones y experimentos para identificar cómo las variables se influyen entre sí. Un desafío central en su desarrollo es la validación del desempeño, ya que esta requiere intervenciones reales en el entorno, las cuales pueden resultar costosas o incluso inviables. Por ello, resulta fundamental contar con entornos sintéticos de evaluación que permitan probar sus capacidades de manera controlada. Esta charla presenta The Causality Game, un entorno diseñado para medir la capacidad de los agentes de descubrir relaciones causales y analizar los costos asociados a dicho proceso.

Salient Object Detection (SOD) methods can locate objects that stand out in an image, assign higher values to their pixels in a saliency map, and binarize the map outputting a predicted segmentation mask. The necessity of abundant annotated data and complex network architectures presents a significant challenge in deep-learning Salient Object Detection (deep SOD) and across the broader deep-learning landscape.  This challenge is particularly acute in medical applications in developing countries with limited computational resources. Recent findings demonstrate that coupling a FLIM encoder with an adaptive decoder creates a flyweight network suitable for SOD, requiring significantly fewer parameters than lightweight models and eliminating the need for backpropagation.  A recent tendency is to investigate pre-trained lightweight models rather than deep neural networks in SOD tasks, coping with applications under limited computational resources. In this context, we have investigated lightweight networks using a methodology named Feature Learning from Image Markers (FLIM), which assumes that the encoder's kernels can be estimated from marker pixels on discriminative regions of a few representative images. Moreover, we Cellular Automata (CA) methods have proven successful in data-scarce scenarios but require proper initialization --- typically through user input, priors, or randomness. We propose a practical intersection of these approaches: using FLIM networks to initialize CA states with expert knowledge without requiring user interaction for each image. By decoding features from each level of a FLIM network, we can initialize multiple CAs simultaneously, creating a multi-level framework. For instance, we have used FLIM for detecting parasites in images. Another very important application regarding microscopy image analysis is related to Oral Epithelial Dysplasia (OED), which is a histopathological diagnosis for potentially malignant lesions with a variable risk of malignant transformation. The conventional grading of OED relies on morphological assessment, a process known for significant inter-observer variability, which limits its prognostic reliability. For coping with this problem, we have used transport-based morphometry applied to a superpixel image.