2020
DOI: 10.3389/frobt.2020.00084
|View full text |Cite
|
Sign up to set email alerts
|

Abstract: Autonomous agents perceive the world through streams of continuous sensori-motor data. Yet, in order to reason and communicate about their environment, agents need to be able to distill meaningful concepts from their raw observations. Most current approaches that bridge between the continuous and symbolic domain are using deep learning techniques. While these approaches often achieve high levels of accuracy, they rely on large amounts of training data, and the resulting models lack transparency, generality, an… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1

Citation Types

0
2
0

Year Published

2020
2020
2024
2024

Publication Types

Select...
3
1
1

Relationship

0
5

Authors

Journals

citations
Cited by 5 publications
(2 citation statements)
references
References 45 publications
0
2
0
Order By: Relevance
“…In more recent years, concept learning entered the domain of practical applications in the rapidly growing field of robotics. The already stated advantages of potentially vast reduction of memory and processing power while improving the effect of the response via association of actions and behaviors to a general class of observable inputs drove development of methods and approaches in practical applications of concept learning [25] however this area is too broad and diverse to attempt to discuss here. Interestingly, while as mentioned earlier, a range of results in generative learning of artificial systems demonstrated that structure emergent in the latent representations created by generative models in the process of unsupervised self-learning can be used as a foundation for learning methods and processes based on distillation of characteristic patterns in the observable environment, they were paralleled very recently by advances in investigation of biologic sensory networks [26], [27] that demonstrated commonality of lowdimensional neural representations in processing sensory information by mammals, including humans.…”
Section: A Related Workmentioning
confidence: 99%
“…In more recent years, concept learning entered the domain of practical applications in the rapidly growing field of robotics. The already stated advantages of potentially vast reduction of memory and processing power while improving the effect of the response via association of actions and behaviors to a general class of observable inputs drove development of methods and approaches in practical applications of concept learning [25] however this area is too broad and diverse to attempt to discuss here. Interestingly, while as mentioned earlier, a range of results in generative learning of artificial systems demonstrated that structure emergent in the latent representations created by generative models in the process of unsupervised self-learning can be used as a foundation for learning methods and processes based on distillation of characteristic patterns in the observable environment, they were paralleled very recently by advances in investigation of biologic sensory networks [26], [27] that demonstrated commonality of lowdimensional neural representations in processing sensory information by mammals, including humans.…”
Section: A Related Workmentioning
confidence: 99%
“…the extension of the lexically fixed expression not the sharpest tool in the box to other semantic fields such as not the crunchiest chip in the bag to the schematic construction not the X-est Y in the Z, it also allows the free combination of multiple constructions and what they call the "appropriate violation of usual constraints" (Van Eecke & Beuls 2018). Yet another large application domain concerns experiments on emergent communication, such as, for example (Nevens et al 2019).…”
Section: Computational Operationalizationsmentioning
confidence: 99%