ColorwAI: Generative Colorways of Textiles through GAN and Diffusion Disentanglement

Andrea Alfarano

Schematization of the ColorwAI framework. We show the disentanglement steps for the two architectures, their latent spaces and the inference procedure. Steps 1 and 4 are different for GANs and Diffusion, while 2 and 3 are common

Schematization of the ColorwAI framework. We show the disentanglement steps for the two architectures, their latent spaces and the inference procedure. Steps 1 and 4 are different for GANs and Diffusion, while 2 and 3 are common

Interpretable World Models in Multimodal LLMs – the aim is to investigate the internal representations of large-scale multimodal LLMs to understand how these models perceive and interpret the world. This project focuses on extracting interpretable concepts from the model’s internal state and shedding light first on its understanding of relationships between different modalities (e.g., images, text, etc.) and second on the underlying structure of its knowledge.
Agentic AI for Art Exploration – here the aim is to develop an Agentic AI application that allows users to explore and interact with large-scale art datasets using natural language. The application integrates the “Segment Anything” model for compositional analysis and features a generative AI-powered chat interface for intuitive information retrieval.