1. INTRODUCTION

1.1. Faint galaxy classifications with HST

The majority of explanations for the faint blue galaxy (FBG) excess observed in deep ground-based images (e.g., Kron 1982, K82 - hereafter references are abbreviated; see bibliography - T88, B92, L93, D94, NW95a) involve Irregular/dwarf populations, or objects with M_B -17.0 mag (using H₀ = 75, q₀ = 0.1 throughout, unless indicated otherwise). The ~ 0".1 FWHM resolution provided by HST's Wide Field Planetary Camera 2 (WFPC2) allows the determination of the sub-kpc morphology of distant galaxies - and hence of the population fractions of different galaxy types - over a wide range of epochs. Gr94, D95a, D95b, G95a, Ca95, O96, A96, and R97 used WFPC2 to study the morphological properties of faint galaxies. D95a showed that the FBG counts for I₈₁₄ 24.5 mag are dominated by late-type/Irregulars, and suggest that neither a steep local luminosity function (LF), nor strong evolution alone can explain the high observed counts (O96). A96 used a 2-parameter space to assign types for I₈₁₄ 25 and reached similar conclusions as D95a. In view of these HST results, the development of a rigorous morphological classification scheme is important to study galaxy formation and evolution in an unbiased manner. One promising approach maps multivariate information to morphological types using an Artificial Neural Network (ANN; cf. O92, O95, N95). This technique was applied to other available WFPC2 fields in B₄₅₀ ( B), since in B the galaxy counts are steepest and the FBG excess will be most pronounced (O96 & Section 2).