AI summary

This paper introduces LSTM, a novel recurrent network algorithm, to solve a hard problem involving distributed, high-precision, continuous-valued representations and long minimal time lags, demonstrating that LSTM can solve non-trivial problems that other recurrent networks cannot.

Abstract

Standard recurrent nets cannot deal with long minimal time lags between relevant signals. Several recent NIPS papers propose alternative methods. We first show: problems used to promote various previous algorithms can be solved more quickly by random weight guessing than by the proposed algorithms. We then use LSTM, our own recent algorithm, to solve a hard problem that can neither be quickly solved by random search nor by any other recurrent net algorithm we are aware of.