Showing papers on "Continuous automaton published in 1972"

Comparison of Expedient and Optima Reinforcement Schemes for Learning Systems

Abstract: Stochastic automata models have been successfully used in the past for modeling learning systems. An automaton model with a variable structure reacts to inputs from a random environment by changing the probabilities of its actions. These changes are carried out using a reinforcement scheme in such a manner that the automaton evolves to a final structure which is satisfactory in some sense. Several reinforcement schemes have been proposed in the literature for updating the structure of automata [1–4]. Most of these are expedient schemes which in the limit yield structures which are better than a device that chooses the actions with equal probabilities irrespective of the environment's response. A few schemes have also been suggested recently which in the limit lead to a continuous selection of a single optimal action as the output of the automaton, when it operates in a stationary environment and are called optimal schemes [5–7]. The question naturally arises as to which of the schemes are to be preferred ...

Some aspects of linear space automata

Abstract: Linear space automaton is introduced as a generalization of probabilistic automaton and its various properties are investigated Linear space automaton has the abilities equivalent to probabilistic automaton but we can treat the former more easily than the latter because we can make use of properties of the linear space, successfully First the solutions are given for the problems of connectivity, state equivalence, reduction and identification of linear space automata Second, the matrix representation of linear space automaton is investigated and the relations between linear space automaton and probabilistic automaton are shown Third, we discuss the closure properties of the family of all real functions on a free semigroup Σ* which are defined by linear space automata and then give a solution to the synthesis problem of linear space automata Finally, some considerations are given to the problems of sets of tapes accepted by la's and also of operations under which the family of all the output functions of la's is not closed

An automaton with brain‐like properties

Abstract: Starting with a Moore‐type automaton the bases for a brain‐like sequential machine are laid down. The problem is considered both at the level of a physical structure and a state structure. The logic is cellular and variable to accommodate learning and generalization. It is shown that this structure can “learn to live” in a consistent environment. Concepts such as recognition and recall of environmental events, short‐term memory, data generation (analogous to speech production) and attention are shown to be natural attributes of the model.

Select properties of the graph model of a general automaton

Abstract: Select properties are presented for a graph model of a general automaton consisting of a processor, environment and time graph. The properties, stated in the form of theorems and corollaries, deal with connectedness, number of points and lines and indegree and outdegree as the model relates to the automaton's sets, functions and characteristics. The properties are illustrated by an example automaton.

On the Existence of a Periodic Analog of a Finite Connected Automaton

Abstract: The solution of the problem of the existence of an analog with periodic-variable structure of a given fixed-structure connected automaton is presented. Using a Boolean approach, a necessary and sufficient condition for the existence of the periodic analog is proven, and an algorithm that enables us to actually find this analog is given. The saving of the memory capacity needed for the construction of a given automaton is the main interest of the existence of a periodic analog.

Behavior of G-Type Automata in a Matrix Game Against Automata with Linear Strategy

Abstract: Among the tremendous variety of environments (i.e., competitive situations) in which a G-type automaton can find itself (the definition of such an automaton and the basic notation was taken from Sragovich [1]), the following case is of special interest. Let M = ‖aij‖ be a numerical matrix with the dimension k × in, and let A be an automaton capable of m moves consisting of the selection of a number from a column of matrix M. In response to a move Yi, when i = 1, …,k, of a G-type automaton and of the j-th move of the A automaton, the former will be fed a number aij whereas the latter (which is defined by a certain rule) will be fed an element of its input alphabet. For the above G-type automaton, the matrix M and the automaton A are a switching environment.