1976–, Computer scientist
Ruslan Salakhutdinov is a Russian-Canadian computer scientist whose 2006 Science paper with Geoffrey Hinton, Reducing the Dimensionality of Data with Neural Networks, demonstrated that stacked restricted Boltzmann machines pre-trained layer by layer could be fine-tuned to outperform PCA at non-linear dimensionality reduction. With Hinton and Yee-Whye Teh he co-authored the contemporaneous A Fast Learning Algorithm for Deep Belief Nets (2006), generally credited with launching the modern deep-learning era.
Salakhutdinov's later work has spanned matrix factorisation, probabilistic deep learning, multimodal models and modular reasoning architectures. He directed Apple's AI research from 2016 to 2020, splitting time with Carnegie Mellon, before returning full-time to CMU. He has supervised many of the leading younger machine-learning researchers of the 2010s.
Video
Related people: Geoffrey Hinton, Yoshua Bengio
Works cited in this book:
- Dropout: A Simple Way to Prevent Neural Networks from Overfitting (2014) (with Nitish Srivastava, Geoffrey E. Hinton, Alex Krizhevsky, Ilya Sutskever)
Discussed in:
- Chapter 9: Neural Networks, Generative Models