International Conference on Modern Artificial Intelligence & Data Science Systems

Abstract

Despite impressive AI developments, artificial general intelligence remains a distant goal. Today’s AIs are specialized and lack the ability to generalize knowledge across domains. To achieve human-like AI, we must understand the brain itself. Neuroscience provides insight into brain structure and function, offering models for intelligent systems.

This talk explores how neuroscience can inform general AI design—and how AI can in turn assist neuroscience. Their intersection creates powerful opportunities for innovation, especially in fields like machine learning, cognitive science, and medical technology.

Abstract

This talk will address some key decisional issues that are necessary for a cognitive and interactive robot, which shares space and tasks with humans. We adopt a constructive approach based on the identification and the effective implementation of individual and collaborative skills.

The system is comprehensive, aiming to support a full set of abilities so that the robot controller can perform human-robot joint actions as collaborative problem solving. These abilities include geometric reasoning, perspective-taking, managing human and robot knowledge in separate cognitive models, and human-aware planning.

We’ll also explore how the robot’s behavior must be pertinent and acceptable to the human, and how that influences the robot’s decision-making, planning, and communication processes.

Abstract

Reinforcement learning is a rapidly growing field within artificial intelligence that trains agents to learn by interacting with their environment.

This talk begins with the core principles of reinforcement learning—state, action, reward, and value functions—before introducing major algorithmic categories: value-based, policy-based, and model-based approaches.

Real-world applications of reinforcement learning will also be showcased, highlighting its power in solving complex problems in robotics, gaming, and autonomous driving.

Abstract

Recently, eXplainable AI (XAI) has gained significant traction in both academia and industry, offering insights into the decision-making processes of black-box machine learning models. This talk introduces a landscape of XAI methods and the strategies for building explainable systems.

Research shows that effective explanations increase user trust—but overwhelming users with too much data can backfire. Thus, parsimony, or the art of delivering just enough explanation, is key. This presentation introduces HAExA, a human-agent explainability framework that provides goal-oriented, parsimonious explanations for remote robotic agents.