With the advent of the data tsunami enabled by cheap sensors, there is a pressing need to make sense of rich data in a streaming manner i.e. with either limited computing power and/or memory. In this talk we will present two different algorithms in supervised and unsupervised machine learning that are sped up or scaled up thanks to the use of optically generated random features. In the case of transfer learning, we will show that the repurposing of a faster deep learning architecture by using optical computing. In the second case, we will show that the combination of a faster algorithm leveraged with optical random projections can become orders of magnitude faster than usual non-parametric methods for a given accuracy.
September 4 @ 11:50
11:50 — 12:30 (40′)

Igor Carron (LightOn)