Projects

Welcome to the forefront of a research journey that embarks on the exciting realm of Pattern Recognition, where we unravel the mysteries of automated pattern detection and uncover the regularities hidden within complex data. This manifesto outlines our vision, which is inspired by the groundbreaking work of renowned scientists, thinkers, and innovators who have paved the way for pattern recognition's diverse applications. From statistical data analysis to machine learning, and from bioinformatics to predictive analytics, our research embraces the versatility of pattern recognition across various domains.

One captivating and challenging avenue of our research is the exploration of non-conventional domains, where objects, such as graphs, sequences, images, and more, reside in unstructured and often non-metric spaces. In these domains, traditional methods fall short, necessitating customized dissimilarity measures to compute the similarity between patterns. As pioneers in this field, we draw inspiration from the works of Richard Hamming, whose pioneering efforts in coding theory laid the foundation for measuring dissimilarity between data patterns, and Alexandre Chorin, who contributed to the study of numerical methods in analyzing complex and irregular data.

Graphs and networks form a crucial component of our non-conventional domains, and we delve into creating innovative dissimilarity measures tailored to compare graphs of varying sizes. In this endeavor, we are inspired by László Lovász, whose contributions to graph theory and combinatorial optimization are invaluable resources for designing meaningful distance metrics for graph-based objects.

One of our primary objectives is to develop novel representation methodologies that enhance classification performance in machine learning applications. For this, we draw upon the visionary work of Geoffrey Hinton and his revolutionary advancements in neural networks, including the development of representation learning techniques like autoencoders and deep belief networks. Yann LeCun's contributions to convolutional neural networks also serve as a beacon guiding us to design powerful and efficient representations for complex and unstructured data.

Furthermore, we are driven by the ambition to leverage the insights from the burgeoning field of bioinformatics, where pattern recognition plays a crucial role in understanding biological data. The work of Anna Tramontano in applying computational methods to study protein structures and interactions inspires us to seek similar data-driven approaches to tackle the challenges in non-conventional domains.

Our manifesto extends a warm invitation to researchers, thinkers, and visionaries alike, encouraging them to join forces in this exhilarating quest. Together, we aim to push the boundaries of pattern recognition and unlock its untapped potential in non-conventional domains. Through interdisciplinary collaboration and the spirit of innovation, we will create cutting-edge methodologies that not only drive academic progress but also find practical applications in industries.

As we embark on this transformative journey, let us honor the legacy of these influential thinkers while carving our path towards a future where pattern recognition transcends the limits of conventionality and pioneers new frontiers of understanding and knowledge. Are you ready to seize this opportunity to be a part of history in the making? Join us now, and together, let us shape the future of pattern recognition and its data-driven applications for non-conventional domains. The possibilities are boundless, and the adventure awaits!