This paper presents a connectionist expert system designed to assist the seicmic interpreter in oil or gas exploration. The system is a learning-based model, not an analytic one which conventional rule-based systems usually belong to. We use multilayered neural network and the back-propagation algorithm as the learning strategy. While learning, a prototype input list of seismic phenomena is fed into the network as input and with an expert's decision together as its desired output, then weights get adjust. After repeated cycles, the knowledge will be distributed all over the network so taht a set of weighted interconnections constitutes an implicit decision criteria. When learning completes, user presents a list of seismic phenomena to this system to get a decision about oil or gas existence.
|Number of pages||3|
|State||Published - 1 Jan 1990|
|Event||1990 Society of Exploration Geophysicists Annual Meeting - San Francisco, United States|
Duration: 23 Sep 1990 → 27 Sep 1990
|Conference||1990 Society of Exploration Geophysicists Annual Meeting|
|Period||23/09/90 → 27/09/90|