Top 5 (or 7) Most Influential Researchers in Artificial Neural Networks

F.Rosenblatt (1928–1971)
For inventing the perceptron - a forerunner of ANNs - and, more importantly, for standing up to the AI charlatans who attached the field in the 1960s in order to drain and divert its funding to AI. Rosenblatt directly confronted M. Minsky and the rest of the AI propaganda machine in various conferences and until his premature death in a boating accident in 1971. His death is by far the greatest loss for the field of ANNs and one of the main reasons that charlatanism prevailed for nearly two decades.

D. H. Hubel (1926-2013) and T. N. Wiesel (1924-present).
For their work on the structure and function of the visual cortex and particularly their discovery of local receptive fields, which led to iterative structures such correlations in ANNs. K. Fukushima and Y. LeCun deserve honorable mention for transferring and popularizing correlation (misnamed as convolution by LeCun) to ANNs but both researches failed to convey (and perhaps comprehend) that the importance is not in the correlation but rather in its iterative and local use. Hubel and Wiesel received the Nobel Prize in Medicine in 1981.

D. E. Rumelhart (1942-2011) J. L. McClelland (1948-present)
For their two volumes Parallel Distributed Processing: Explorations in the Microstructure of Cognition and the related work, which revived the field of neural networks (and funding for it) in the 1980s. Gradient descent learning for neural networks had been previously discovered in various forms and by various researchers during the 1960s and 1970s. However, its version in the PDP volumes, named back propagation, and its various successful applications registered with the Computer Science community and funding institutions and dispelled the broadly accepted myths and prejudice that AI propaganda machine had generated and perpetuated for nearly two decades.
J. Schmidhuber (1963-present)
For his early and visionary work in deep neural networks, game-theoretic learning, recurrent networks, and the invention of LSTM units. His work on adversarial learning in 1992 was rediscovered, renamed and popularized as GANs in 2014. Schmidhuber is a visionary researcher and a true leader of the field.

S. Hochreiter (1967-present)
For his early and visionary work in deep neural networks, recurrent networks, the invention of LSTM and SELU units, and self-normalizing neural networks. Hochreiter is one of the most sophisticated and most active researchers in the field.