Over 400 participate in the III Edition of the ValgrAI Scientific Council Forum
Valencia, July 18, 2024.- The III ValgrAI Scientific Council Forum, promoted by the ValgrAI Foundation, brought together more than 400 people at the Nexus Building of the Polytechnic University of Valencia (UPV), attracted by the comprehensive international program of AI experts.
The event, led by the Managing Director of ValgrAI, Ana Cidad, featured the presence of the General Director of ValgrAI, Vicent Botti; the General Director of Science and Research of the Generalitat Valenciana, Rafael Sebastián: and the Vice-Rector of Research at UPV, Belén Picó, who highlighted the transformative power of Artificial Intelligence (AI) and the boost that AI research being developed in the Valencian Community represents for numerous sectors.
Following them, one of the most anticipated presentations was given by Hiroaki Kitano, CEO of SonyAI and Vice President of Sony Corporation, who, under the title “Nobel Turing Challenge: Creating the Engine for Scientific Discovery” emphasized the importance of accelerating the pace of scientific discoveries, stating that this effort requires not only a deep understanding of the areas of study but also of the very nature of scientific discoveries. According to Hiroaki Kitano, it is essential to establish a “science of science” that can be practically implemented through AI systems. Hiroaki Kitano explained that AI “scientists” might not resemble human scientists in their processes and that this alternative form of science could overcome the cognitive and sociological limitations that currently restrict scientific practice. This approach could lead to a hybrid form of science, combining human and AI efforts, taking disciplines such as systems biology and other sciences to a new stage.
Therefore, the “Nobel Turing Challenge” project aims to develop a highly autonomous AI system capable of performing top-tier science, indistinguishable from the quality produced by the best human scientists. This challenge seeks for some of the discoveries made by these AIs to be worthy of a Nobel Prize. With this vision, Hiroaki Kitano envisions a future where collaboration between humans and machines will not only enhance the pace of scientific discoveries but also redefine the very nature of scientific research.
Subsequently, Ernesto Faubel from LDT CitiVERSE EDIC and the City Council of Valencia, along with Patricia Tamarit from Nunsys, presented the project “AI models for smart cities – EDIC and European Local Digital Twin toolbox”, which aims to create digital twins of major cities and buildings in Europe, starting with pilot cities where simulations can be developed and conducted to help predict future situations. Specifically, Valencia is the representative city of Spain in this European initiative.
The Future of AI
Under the title “The future of AI: ValgrAI Researches,” some ValgrAI researchers presented various projects that demonstrate the transformative impact of artificial intelligence and robotics in fields such as health, security, and information accessibility.
- Artificial Intelligence in Video Games: Carlos Marín Lora from Universitat Jaume I and a member of the GAMERS – UJI research group presented his study on how AI techniques can transform the video game experience, making them more immersive and personalized. His advancements range from active games for treadmills to cultural promotion through geolocation and augmented reality.
- Autonomous Robotics: Antonio Santo from the Miguel Hernández University (Elche) demonstrated a method for navigating autonomous robots in natural environments using 3D point clouds obtained with LiDAR sensors, improving their ability to identify traversable areas.
- Prostate Cancer Diagnosis: Alejandro Golfe San Martín from the Polytechnic University of Valencia, along with the CVBLab group, presented an improved content-based image retrieval (CBIR) system using AI-generated synthetic samples, increasing the accuracy of prostate cancer diagnosis.
- Home Autonomy for Dependent People: Isabel Ferri Molla from the Polytechnic University of Valencia presented a multimodal interaction system that uses neural networks and augmented reality to assist elderly and dependent people in their daily tasks, enhancing their autonomy.
- Automatic Translation of Historical Documents: Miguel Domingo from the Polytechnic University of Valencia showcased advancements in applying modern language to make historical documents accessible, facilitating cultural heritage preservation.
- Adaptation of Vision-Language Models in Digital Pathology: Pablo Meseguer Esbri from the Polytechnic University of Valencia explained an adaptation method without training for classifying histopathological images, improving accuracy and speed in cancer diagnoses.
- Forecasting Extreme Climatic Events: Oscar Pellicer from the University of Valencia developed an LSTM convolution model to predict impacts on vegetation and provide explainable analyses of extreme climatic events, offering early warning signals.
- Anomaly Detection in the Internet of Things: Sahibzada Saadoon Hammad from Universitat Jaume I presented research on detecting anomalies in IoT devices using Edge Computing, highlighting the benefits and challenges of this analysis.
- Underwater Robotics for Inspection and Maintenance: Salvador López Barajas from Universitat Jaume I proposed a robotic platform for inspecting and repairing underwater structures, essential for industries such as aquaculture and mining.
- Mitigating Bias in Language Models for Text Simplification: Victoria Muñoz García from the University of Alicante addressed the mitigation of bias in language models used for simplifying complex texts, improving their accessibility.
The morning session was concluded by Carme Torras, a researcher at CSIC, with the presentation “Social robotics: Research challenges and ethics education initiatives”, where she addressed the growing field of assistive robotics, highlighting its importance in aiding healthcare professionals in hospitals, rehabilitation centers, and nursing homes, as well as empowering people with reduced mobility in their homes to perform their daily activities autonomously.
Torras emphasized that operating in dynamic, people-centered environments presents new research challenges. Robotic assistants must have user-friendly interfaces, be highly adaptable and customizable, extremely safe for people, and capable of handling deformable materials. These technical requirements are fundamental to ensuring that robots can integrate effectively and safely into human environments.
In addition to the technical challenges, assistive robotics also poses significant ethical challenges, leading to the emergence of a new discipline: Roboethics. Torras mentioned that several institutions are developing regulations and standards to address these ethical challenges. Additionally, numerous ethics education initiatives include content on human-robot interaction and human dignity in assistance situations.
In the afternoon session, the focus shifted to the business sector, where various companies presented their success cases in the application of Artificial Intelligence, as well as the situations they face daily in its use. Thus, Ana Isabel Prieto from R&D at S2 Grupo gave the presentation “Unmasking Spoofers: navigating safely thanks to AI”; Pablo Alcoriza, CTO at Ídrica, presented “GoAigua: Smart Water Solutions”; Javier Anguiano Aranzubía, Delegate of Cetaqua in CV, talked about “AI as a transforming element in the water sector”; Víctor Peris, CEO of INDAWS, shared his experience “Applying AI to business management”; Noelia González Méndez, Manager at NTT DATA, and Rubén Almaraz Arranz, Project Leader at NTT DATA, presented “AI-Based Processes: Artificial Intelligence applied to business process optimization”; Manuel Herranz, CEO of Pangeanic, focused on “GenAI that works: Pangeanic at the EFE Agency and the Tax Agency” and Jorge Capel, head of the solution architecture office at Nunsys, addressed “Expert knowledge through Generative AI.”
The closing presentation was delivered by Professor Emeritus of Oregon State University, Thomas Dietterich, titled “Integrating machine learning into safety-critical systems”, where he emphasized the capabilities of systems created through deep learning and their application in safety-critical areas, such as medicine, aeronautics, and autonomous cars. The talk focused on how machine learning methodologies are evolving to operate in systems where safety is paramount.
Dietterich detailed several key changes necessary to adapt machine learning to these high-demand environments, specifically the construction of high-fidelity simulators, replicating situations accurately in controlled conditions, adversarial data collection for training, and specifically, of dangerous regions within the operational design domain (ODD), ensuring the system is prepared to handle risk situations.
He also stressed that it is crucial for models to adjust correctly to new situations, verifying their performance beyond the training data; and that methods be developed to calculate the probability of failures and damages under normal operating conditions to maintain system safety.
Dietterich also pointed out that there are many research challenges to achieving these goals and emphasized that traditional safety engineering only addresses known hazards, so it is crucial to design through Artificial Intelligence systems that also detect and respond to new hazards.
