This is an old revision of the document!

AUTONOMY & MEANING

Three Pillars of Intelligence: Control, Classification, Learning

Control	Is the ability to act with a purpose - affect the world in a way that has a target to be achieved. Anything from a thermostat controlling the temperature in a room to a human trying become rich and famous are examples of control systems. In the former case, measuring when the goal is achieved is easier than in the latter.
Classification	To act, one needs to know what to act on. To know what to act on, one needs to classify. To classify, one needs to sense. To sense, one needs measurement devices.
Learning	… is the accumulation of information for the purpose of control and classification. In a world that cannot be perceived all-at-once yet contains regularity, learning is necessary to conserve energy.

Control

What it is	Systematic production of an effect in a world, through prolonged commitment to actuation based on measurement, in light of a goal (end state).
Measurement	Recording and storage of values of particular variables over time.
Effect	Realization of a causal relation in the (physical) world.
Transducer	A device that changes one type of energy to another, typically amplifying and/or dampening the energy in the process.
Sensor	A transducer that changes one type of energy to another type.
Action	An effect that a controller can inflict on its external environment.
Actuator	A physical (or virtual/simulated) transduction mechanism that implements an action that a controller has committed to (e.g. gripper/hand).
Control Connection	Causal connection between a control signal and an actuator.
Adaptation	Modification of a control connection in light of a goal g that has not been achieved. Requires the control connection structure to change.
Digital Controller	Separates the stages of measurement, analysis, and control. Makes adaptive control in machines more feasible than a mechanical connection.
Feedback	For a variable v, information of its value at time t1 is transmitted back to the controller through a feedback mechanism as v', where v'(t) > v(t) that is, there is a latency in the transmission, which is a function of the speed of transmission (encoding (measurement) time + transmission time + decoding (read-back) time).
Latency	A measure for the size of the difference between v and v'.
Jitter	The change in Latency over time. Second-order latency.

Classification

What it is	Separation of some signals or patterns from other signals or patterns.
Example	Artificial Neural Networks (ANNs). Contemporary ANNs (e.g. Deep Neural Networks, Double-Deep Q-Learners, etc.) can only do classification. They can only do classification by going through a long continuous training session, after which the learning is turned off.
Example of misguided use of classification	download video
Unified Control & Classification	To be an intelligent agent, classification must be unified with control and learning to produce am agent that can control its classification as turns out necessary to learn.

Learning

What it is	Learning is a process that has the intent of acquiring actionable information, a.k.a. knowledge.
Key Features	Inherits key features of any process: - Purpose: To adapt, to respond in rational ways to problems / to achieve foreseen goals; this factor determines how the rest of the features in this list are measured. - Speed: The speed of learning. - Data: The data that the learning (and particular measured speed of learning) requires. - Quality: How well something is learned. - Retention: The robustness of what has been learned - how well it stays intact over time. - Transfer: How general the learning is, how broadly what is learned can be employed for the purposes of adaptation or achievement of goals. - Meta-Learning: A learner may improve its learning abilities - i.e. capable of meta-learning. - Progress Signal(s): A learner needs to know how its learning is going, and if there is improvement, how much.
Measurements	To know any of the above some parameters have to be measured: All of the above factors can be measured in many ways.
Major Caveat	Since learning interacts with (is affect by) the task-environment and world that the learning takes place in, as well as the nature of these in the learner's subsequent deployment, none of the above features can be assessed by looking only at the learner. This is addressed by the Pedagogical Pentagon.

Learning Controllers

A Learner

Adaptive/intelligent system/controller, embodied and situated in a task-environment, that continually receives inputs/observations (measurements) from its environment and sends outputs/actions back (signals to its manipulators).
Some of the learner’s inputs may be treated specially — e.g. as feedback or a reward signal, possibly provided by a teacher or a specially-rigged training task-environment. Since action can only be evaluated as “intelligent” in light of what it is trying to achieve - we model intelligent agents as imperfect optimizers of some (possibly unknown) real-valued objective function.
Note that this working definition fits experience-based learning.

Embodiment

The interface between a learning controller and the task-environment.

Unification of Control, Classification & Learning

Learning to Classify	In a world with a lot of variation, a learning controller must also learn to classify. To learn to classify, one must learn what to control to classify appropriately. Learning to control, therefore, requires learning two kinds of classification (at the very least).

Meaning

Meaning	The meaning of a datum d to an agent A is defined by the effect that d has on the behavior of A.
More Specifically	Meaning is a mental state of an agent A that affects the control function of an autonomous agent. Given an agent A with a set of differentiable goals G, knowledge K, and a statement, S, the meaning of S is defined by how it selects between goals g in G, by preventing some and enabling other elements in G to be pursued.
Producing Meaning	Meaning is produced through a process of understanding using reasoning over causal relations.
Causal Relations	The relationship between two or more differentiable events such that one of them can (reasonably reliably) produce the other. One event E, the cause, must come before another event E', the effect, where E can (reasonably reliably) be used to produce E'.

Understanding

To consistently solve problems regarding a phenomenon <m>X</m> requires understanding <m>X</m>.
Understanding <m>X</m> means the ability to extract and analyze the meaning of any phenomena <m>\phi</m> related to <m>X</m>.

Bottom Line

Can't talk about creativity without talking about understanding, and can't talk about understanding without talking about meaning. No good scientific theory of meaning (but some philosophical ones) exists.

Autonomy

Center for Analysis and Design of Intelligent Agents

Table of Contents