Connectionism
One of the principal research paradigms in artificial intelligence, represented by neural networks and deep learning. It relies on gradient descent to optimize across large parameter spaces and holds that intelligence emerges from the connections and interactions among large numbers of simple processing units.
Core Facts
Timeline (last 90 days)
Symbolic systems generalize through rules, while connectionist models generalize through statistical similarity.
One advantage of logical complexity is transparency and interpretability, whereas the difficulty readings of neural networks are hard to reverse-engineer to reveal the internal structure of a concept.
Neural networks are not the antithesis of symbolic reasoning, but rather an 'emergent implementation' of symbolic reasoning.
The debate between connectionism (neural networks) and symbolism (classical AI) has persisted for decades.
Symbolism holds that the essence of intelligence is symbol manipulation and logical reasoning, while connectionism holds that intelligence emerges from the parallel connection and learning of large numbers of simple units.
All Facts (6)
The McCulloch–Pitts neuron model was proposed in 1943 and is the origin of the connectionist approach
65%UnverifiedSymbolic systems generalize through rules, while connectionist models generalize through statistical similarity.
50%UnverifiedOne advantage of logical complexity is transparency and interpretability, whereas the difficulty readings of neural networks are hard to reverse-engineer to reveal the internal structure of a concept.
50%UnverifiedNeural networks are not the antithesis of symbolic reasoning, but rather an 'emergent implementation' of symbolic reasoning.
50%UnverifiedThe debate between connectionism (neural networks) and symbolism (classical AI) has persisted for decades.
50%UnverifiedSymbolism holds that the essence of intelligence is symbol manipulation and logical reasoning, while connectionism holds that intelligence emerges from the parallel connection and learning of large numbers of simple units.
50%