Cisco Systems, Inc.
Neural network-assisted computer network management

Abstract:

Sequences of computer network log entries indicative of a cause of an event described in a first type of entry are identified by training a long short-term memory (LSTM) neural network to detect computer network log entries of a first type. The network is characterized by a plurality of ordered cells F.sub.i=(x.sub.i, c.sub.i-1, h.sub.i-1) and a final sigmoid layer characterized by a weight vector w.sup.T. A sequence of log entries x.sub.i is received. An h.sub.i for each entry is determined using the trained F.sub.i. A value of gating function G.sub.i(h.sub.i, h.sub.i-1)=II (w.sup.T(h.sub.i-h.sub.i-1)+b) is determined for each entry. II is an indicator function, b is a bias parameter. A sub-sequence of x.sub.i corresponding to G.sub.i(h.sub.i, h.sub.i-1)=1 is output as a sequence of entries indicative of a cause of an event described in a log entry of the first type.