Cloudy with a chance of neurons: The tools that make NNs work In an earlier DL article, we talked about how inference workloads—
the use of already-trained NNs to analyze data—can run on fairly cheap hardware, but running the training workload that the NN “learns” on is orders of magnitude more expensive.
In particular, the more potential inputs you have to an algorithm, the more out of control your scaling problem gets when analyzing its problem space.
This is where MACH, a research project authored by Rice University’s Tharun Medini and Anshumali Shrivastava, comes in.
MACH is an acronym for Merged Average Classifiers via Hashing, and according to lead researcher Shrivastava, “[its] training times are about 7-10 times faster, and… memory footprints are 2-4 times smaller” than those of previous large-scale DL techniques.
In describing the scale of extreme classification problems, Medini refers to online shopping search queries, noting that “there are easily more than 100 million products online.
If this article appears incomplete, it may be intentional. Try prompting for a continuation.