The document compares the performance of an autoassociative memory with and without using a pseudoinverse weight matrix. It finds that using the pseudoinverse weight matrix improves performance in both noise-free conditions and when noise is present. Specifically, it finds that without the pseudoinverse, the weight matrix has a larger range of values and more cross-correlation, resulting in more character errors. With the pseudoinverse, the weight matrix range is limited to 0 to 1, improving performance both without and with noise. The autoassociative memory using the pseudoinverse weight matrix thus demonstrates much better performance.