A linear-bounded automaton (LBA) is a type of theoretical computing machine that operates within a limited amount of memory. It is a variant of a Turing machine where the tape size is restricted to a linear function of the input size. This means that the LBA can only use memory proportional to the length of the input string, making it more constrained than a standard Turing machine. LBAs are used to recognize a specific class of languages known as context-sensitive languages. These languages are more complex than context-free languages but less complex than those recognized by unrestricted Turing machines. The linear memory limitation allows LBAs to perform computations efficiently while still being powerful enough to handle certain types of problems.