Abstract:Autonomous lifelong development and learning is a fundamental capability of humans, differentiating them from current deep learning systems. However, other branches of artificial intelligence have designed crucial ingredients towards autonomous learning: curiosity and intrinsic motivation, social learning and natural interaction with peers, and embodiment. These mechanisms guide exploration and autonomous choice of goals, and integrating them with deep learning opens stimulating perspectives. Deep learning (DL) approaches made great advances in artificial intelligence, but are still far away from human learning. As argued convincingly by Lake et al., differences include human capabilities to learn causal models of the world from very little data, leveraging compositional representations and priors like intuitive physics and psychology. However, there are other fundamental differences between current DL systems and human learning, as well as technical ingredients to fill this gap, that are either superficially, or not adequately, discussed by Lake et al. These fundamental mechanisms relate to autonomous development and learning. They are bound to play a central role in artificial intelligence in the future. Current DL systems require engineers to manually specify a task-specific objective function for every new task, and learn through off-line processing of large training databases. On the contrary, humans learn autonomously open-ended repertoires of skills, deciding for themselves which goals to pursue or value, and which skills to explore, driven by intrinsic motivation/curiosity and social learning through natural interaction with peers. Such learning processes are incremental, online, and progressive. Human child development involves a progressive increase of complexity in a curriculum of learning where skills are explored, acquired, and built on each other, through particular ordering and timing. Finally, human learning happens in the physical world, and through bodily and physical experimentation, under severe constraints on energy, time, and computational resources. In the two last decades, the field of Developmental and Cognitive Robotics (Cangelosi and Schlesinger, 2015, Asada et al., 2009), in strong interaction with developmental psychology and neuroscience, has achieved significant advances in computational

Computational Theories of Curiosity-Driven Learning

Understanding exploration in humans and machines by formalizing the function of curiosity

From Psychological Curiosity to Artificial Curiosity: Curiosity-Driven Learning in Artificial Intelligence Tasks

Meta-learning curiosity algorithms

Curiosity as a Metacognitive Feeling

Curiosity and the dynamics of optimal exploration

Curious Minds Wonder Alike: Studying Multimodal Behavioral Dynamics to Design Social Scaffolding of Curiosity

A transdisciplinary view on curiosity beyond linguistic humans: animals, infants, and artificial intelligence

Prototyping three key properties of specific curiosity in computational reinforcement learning

Dynamic Neural Curiosity Enhances Learning Flexibility for Autonomous Goal Discovery

Five Properties of Specific Curiosity You Didn't Know Curious Machines Should Have

The Psychology and Neuroscience of Curiosity

The growth and form of knowledge networks by kinesthetic curiosity

A Novel Multimodal Approach for Studying the Dynamics of Curiosity in Small Group Learning

Computational Curiosity (A Book Draft)

A New Theoretical Framework for Curiosity for Learning in Social Contexts

Mobile Robot’s Sensorimotor Developmental Learning from Orientation and Curiosity

Novelty and habituation: the driving forces in early stage learning for developmental robotics

Autonomous development and learning in artificial intelligence and robotics: Scaling up deep learning to human--like learning

Computational mechanisms of curiosity and goal-directed exploration

CURIOUS: Intrinsically Motivated Modular Multi-Goal Reinforcement Learning