Lev Goldfarb  obtained Diploma (~ MSc) in Mathematics & Computer Science (specializing in topology) from St.-Petersburg State University, Russia, and Ph.D. in Systems Design Engineering (specializing in pattern recognition) from the University of Waterloo, Canada.  He was then awarded Natural Sciences and Engineering Research Council of Canada (NSERC) Postdoctoral Fellowship for 1980–82.  The following twenty five years he worked as an Assistant and then Associate Professor in the Faculty of Computer Science, University of New Brunswick, Fredericton, Canada.  After early retirement, he now conducts research and consulting through company Inductive Information Systems (IIS).


He has served on the editorial boards of Pattern Recognition, Pattern Recognition Letters, Cognitive Neurodynamics.


Trained as a mathematician, he realized at the beginning of his career the fundamental inadequacy of the conventional pattern recognition (especially statistical) formalisms, and so all professional life he has been working on the development of new formalisms that would serve more adequately the needs of the field.  In the early 80s, he pioneered an approach to the problem of efficient classification of the conventionally structured (non-vector) data via its near-isometric representation in the corresponding low-dimensional pseudo-Euclidean vector space.  Over the last decade and a half, this approach has received renewed interest in image retrieval, pattern recognition, and machine learning.


Starting from the late 1980s and prompted by the true novelty and variety of geometries he observed within a symbolic representation—as compared to the conventional mathematical spaces—and also by the intrinsic inability of the latter to support an adequate concept of object class, he, with his students, have been developing a fundamentally new kind of representational formalism (ETS).  It appears that this formalism for structural representation has much wider scientific implications.