Transparent Neural Networks: Integrating Concept Formation and Reasoning

• We present the transparent neural networks, a graph-basedcomputational model that was designed with the aim of facilitating humanunderstanding. We also give an algorithm for developing such networksautomatically by interacting with the environment. This is doneby adding and removing structures for spatial and temporal memory.Thus we automatically obtain a monolithic computational model whichintegrates concept formation with deductive, inductive, and abductivereasoning.

Subject headings

HUMANIORA  -- Filosofi, etik och religion -- Filosofi (hsv//swe)

HUMANITIES  -- Philosophy, Ethics and Religion -- Philosophy (hsv//eng)

HUMANIORA  -- Filosofi, etik och religion (hsv//swe)

HUMANITIES  -- Philosophy, Ethics and Religion (hsv//eng)

NATURVETENSKAP  -- Data- och informationsvetenskap -- Datavetenskap (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences -- Computer Sciences (hsv//eng)

NATURVETENSKAP  -- Data- och informationsvetenskap -- Datorseende och robotik (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences -- Computer Vision and Robotics (hsv//eng)

Keyword

transparent neural networks

inductive reasoning

developmental robotics

deductive reasoning

concept formation

concept formation; deductive reasoning; developmental robotics; inductive reasoning; transparent neural networks

Publication and Content Type

kon (subject category)

ref (subject category)