THE ROLE OF WEIGHT NORMALIZATION IN COMPETITIVE LEARNING
THE ROLE OF WEIGHT NORMALIZATION IN COMPETITIVE LEARNING
复制标题
DOI：
10.1162/neco.1994.6.2.255
复制
发表时间：
1994-03-01
影响因子：
2.9
通讯作者：
BARROW, HG
中科院分区：
文献类型：
--
作者：
GOODHILL, GJ;BARROW, HG
分析了不同类型的权重归一化对一个简单的竞争学习规则的结果的影响。它表明，有重要的差异，在所形成的表示取决于是否强制执行的约束是由相同的量除以每个权重（“分裂的强制执行”）或减去一个固定的量从每个权重（“减法强制执行”）。对于分裂的情况，权重向量在空间上展开，以便均匀地表示“典型”输入，而对于减法的情况，权重向量倾向于空间的轴，以便表示“极端”输入。这些差异的后果进行检查。
The effect of different kinds of weight normalization on the outcome of a simple competitive learning rule is analyzed. It is shown that there are important differences in the representation formed depending on whether the constraint is enforced by dividing each weight by the same amount (''divisive enforcement'') or subtracting a fixed amount from each weight (''subtractive enforcement''). For the divisive cases weight vectors spread out over the space so as to evenly represent ''typical'' inputs, whereas for the subtractive cases the weight vectors tend to the axes of the space, so as to represent ''extreme'' inputs. The consequences of these differences are examined.