When the subject of Artificial Neural Networks took off in the
late 1980s, researchers hoped that a system of artificial
intelligence directly inspired by biological nervous systems would
help them understand the working of the human brain. Since then,
these information processing systems, which are based on large
numbers of highly interconnected, simple processors, have made their
mark in applications that first train them on large databases of
prior examples and then use them to detect complex trends or
patterns in another body of data.
The Special Topics 10-year ranking of the hottest papers reflects
the applications where neural networks are now put to use. The list
leads off with a paper on the use of neural networks for the
classification and diagnostic prediction of cancers. A quartet of
papers describes the use of these networks for predicting the
characteristics of proteins. Other applications in the top 20 list
include face detection, gene expression patterns, forecasting water
resource use, soil hydraulics, and atmospheric sciences. Seven of
the hottest papers from the past decade are analyses of the problem
of global stability for networks with time delays.
It’s this question of global stability in neural networks that
far and away dominates the ranking of the hottest papers from the
last two-years. Fifteen of the top 20 papers examine this one issue,
with applications in proteomics and genomics and reviews filling out
the remaining five.
Methodology
To construct this database,
papers were extracted based on title-supplied keywords for Artificial
Neural Networks. The keywords used were as follows:
"neural network*"
NOT "central*" NOT "CNS" NOT "*brain*" NOT "neuron*" NOT "cortical"
The baseline time span for this database
is 1997-August 31, 2007 (fourth bimonthly period in 2007). The resulting database contained
19,391 (10 years)
and 6,156 (2 years) papers; 31,861 authors;
118 countries; 2,448 journals; and 6,963 institutions.
Rankings
Once the database was in place,
it was used to generate the lists of top 20 papers (two- and ten-year
periods), authors, journals,
institutions, and nations, covering a time span of 1997-August 31, 2007 (fourth bimonthly
period of 2007, a 10-year plus 8-month period).
The top 20 papers are ranked
according to total cites. Rankings for author, journal, institution,
and country are listed in three ways: according to total cites, total
papers, and total cites/paper. The paper thresholds and corresponding
percentages used to determine
scientist, institution, country, and journal rankings according to
total cites/paper, and total papers respectively are as follows:
| Entity: |
Scientists |
Institutions |
Countries |
Journals |
| Thresholds: |
12 |
53 |
25 |
15 |
| Percentage: |
1% |
1% |
50% |
10% |
