7 - Symbolism vs. Connectionism: A Closing Gap in Artificial Intelligence -

7 - Symbolism vs. Connectionism: A Closing Gap in Artificial Intelligence -

Wang (2017)

Dit paper bespreekt symbolisme en connectionisme:

•Wat zijn symbolische en connectionistische AI?•Wat zijn de voordelen van symbolische AI? EN van connectionistische AI?•Hoe beargumenteert de auteur dat het onderscheid tussen symbolische en connectionistische AI steeds verder vervaagt of zelfs verdwijnt?

• AI: “find how to make machines use language, form abstractions and concepts, solve kinds of

problems now reserved for humans, and improve themselves.” (1956) → focused on the symbolic capacities

- central problem of AI: how knowledge is represented, encoded, and processed

- main debate: the dichotomy (contrast) of symbolic and connectionist paradigms

• AGI: artificial general intelligence; human level AI that is capable of completing a wide range

of tasks in an appropriate fashion

• symbolic AI was conceived in the attempt to explicitly represent human knowledge in facts,

rules, and other declarative, symbolic forms

• symbol: a pattern that stands for other things (target can be object, symbol or relation)
• the nature of human language is the organization of signs

→ individual signs have limited ability to convey meanings unless embodied in a sign system

• by abstracting symbols from lower levels to higher levels (physical, cognitive, social,

narrative), we form abstract concepts and find universal meanings

• physical symbol system hypothesis (PSSH): the PSS has the necessary and sufficient

means for general intelligent action, it is the only way toward AGI

• physical symbol system (PSS): a physical computing device for symbol manipulation, which

consists of discrete symbols

• symbols: form expressions, or symbol structures through some sort of physical connections
• physical structures often work as internal representations of the environments and also contain

a set of processes that “operate on expressions to produce other expressions” → PSSH implies that the existence of symbolic-level computing in a system is independent of the physical substrate it operates on

• representation: the mapping from one sign system to another (semiotic morphism)
• knowledge representation: to represent information about the world in a system in a way the

system can employ to store and retrieve old information, infer new knowledge, and perform complex functions (main problem in AI) 1 / 2

• symbolic approaches represent knowledge in a highly structured fashion
• the basic units of symbolic representation are symbolic atoms, specific words or concepts

→ this representation paradigm is also called localist as opposed to distributed representation in connectionist models

• existential graph (EG): symbolic in the sense that it uses individual nodes and arcs to represent

different concepts and their relationships, it captures the aggregate structures of knowledge

• semantic networks: use graphic notations to represent individual objects and categories of

objects → the notations include nodes that are connected by labeled links, which represent relations among objects

• symbolic paradigm is criticized for many reasons, for example: many symbolic structures need

manual coding → it is believed that an essential component of intelligence is a physical body that interacts with the environment through perceptions and behaviors, “grounding” the symbols to the world and giving them meanings (purely manipulating symbols misses that) → intelligence requires non-symbolic processing (but the relationship between symbolic and non-symbolic prodess is supplement instead of replacement) → computers will never be the same as brains (but they both involve computation processes, both brains and computers are essentially physical symbol systems that can give rise to intelligence)

• representation-transformation: focuses on information process (instead of computers)

→ Boolean dream: problems explored in symbolic paradigm are too simple from a neural science point of view, unable to provide rich insight for the computational organization of the brain

• rules can always lead from true statements to other true statements and see thinking

as the manipulations of propositions → symbolic models are only succesful at coarse levels, unable to account for the detailed structure of cognition

• connectionist models refer to bio-inspired networks composed of a large number of

homogenous units and weighted connections among them, analogous to neurons and synapses in the brain → the strengths of the connections reflect how closely the units are linked and can be strengthened or weakened dynamically by new training data → the main task of connectionist paradigm is to tune the weights until the optimum is reached through techniques like gradient descent

• the signals in the neural nets in brains can be modeled by logic expressions , exhibiting digital

properties

• / 2