principal components method, the template for each neuron is made of a few coefficients obtained by projection. The advantage of the principal components method is operation in a much lower dimensional space. However, some preprocessing of data is required in the principal components method. The principal components melhod, also known as the Karhunen-Loeve Transform (KL T) in the signal processing field represents the data in a transformed space spanned by a set of orthonormal basis vectors derived from the data. Let each waveform be represented by an N dimensional column vector x, the covariance matrix of the waveform vectors is: m)'}, (1)

DOI link for principal components method, the template for each neuron is made of a few coefficients obtained by projection. The advantage of the principal components method is operation in a much lower dimensional space. However, some preprocessing of data is required in the principal components method. The principal components melhod, also known as the Karhunen-Loeve Transform (KL T) in the signal processing field represents the data in a transformed space spanned by a set of orthonormal basis vectors derived from the data. Let each waveform be represented by an N dimensional column vector x, the covariance matrix of the waveform vectors is: m)'}, (1)

principal components method, the template for each neuron is made of a few coefficients obtained by projection. The advantage of the principal components method is operation in a much lower dimensional space. However, some preprocessing of data is required in the principal components method. The principal components melhod, also known as the Karhunen-Loeve Transform (KL T) in the signal processing field represents the data in a transformed space spanned by a set of orthonormal basis vectors derived from the data. Let each waveform be represented by an N dimensional column vector x, the covariance matrix of the waveform vectors is: m)'}, (1)