Endowing machines with the abilities to represent and understand natural language as well as to ground it to the physical world is a longstanding challenge for the development of AI systems simulating human cognitive and communicative abilities. Last years have seen a surge of interest in AI personal assistants and humanoid robots, able to converse with humans, take decisions and execute actions on their behalf. However, whilst these AI systems are improving rapidly to recognize speech, their ability to understand natural language and communicate using it with humans is still limited. Indeed, even if they may appear more competent with respect to the past, they suffer from the limitations of being mostly rule-based, not generalizing or scaling beyond their programmed domains and, most importantly, not learning and using robust and meaningful representations of physical concepts, objects and events.
This Research Topic is intended to provide an overview of the research being carried out in the areas of cognitive and AI systems designed to learn capabilities for language learning, understanding and grounding. As language learning and grounding problem requires an interdisciplinary attitude, the research topic aims to gather researchers with broad expertise in various fields — machine learning, computer vision, natural language, neuroscience, and psychology — to discuss their cutting edge work as well as perspectives on future directions in this exciting space of language, grounding and interactions. Therefore, this collection aims to address the following problems:
• How natural language can be represented into an artificial system
• How to jointly represent verbal and visual information coming from different perceptual systems
• How to store, selectively process and form words and sentences in natural language tasks
• How to ground words in perceptual representations of the visible surroundings and embodied experience
• How to answer questions emulating natural language reasoning
• How to learn and progressively improve communicative and multimodal skills, interactively or autonomously
• How to build an internal representation of the world and effectively reuse it to address novel and unknown tasks
• How to detect sentiments and emotions in language expressions.
Original contributions addressing these issues are sought, covering the whole range of theoretical and practical aspects, technologies and systems.
Endowing machines with the abilities to represent and understand natural language as well as to ground it to the physical world is a longstanding challenge for the development of AI systems simulating human cognitive and communicative abilities. Last years have seen a surge of interest in AI personal assistants and humanoid robots, able to converse with humans, take decisions and execute actions on their behalf. However, whilst these AI systems are improving rapidly to recognize speech, their ability to understand natural language and communicate using it with humans is still limited. Indeed, even if they may appear more competent with respect to the past, they suffer from the limitations of being mostly rule-based, not generalizing or scaling beyond their programmed domains and, most importantly, not learning and using robust and meaningful representations of physical concepts, objects and events.
This Research Topic is intended to provide an overview of the research being carried out in the areas of cognitive and AI systems designed to learn capabilities for language learning, understanding and grounding. As language learning and grounding problem requires an interdisciplinary attitude, the research topic aims to gather researchers with broad expertise in various fields — machine learning, computer vision, natural language, neuroscience, and psychology — to discuss their cutting edge work as well as perspectives on future directions in this exciting space of language, grounding and interactions. Therefore, this collection aims to address the following problems:
• How natural language can be represented into an artificial system
• How to jointly represent verbal and visual information coming from different perceptual systems
• How to store, selectively process and form words and sentences in natural language tasks
• How to ground words in perceptual representations of the visible surroundings and embodied experience
• How to answer questions emulating natural language reasoning
• How to learn and progressively improve communicative and multimodal skills, interactively or autonomously
• How to build an internal representation of the world and effectively reuse it to address novel and unknown tasks
• How to detect sentiments and emotions in language expressions.
Original contributions addressing these issues are sought, covering the whole range of theoretical and practical aspects, technologies and systems.