In order to find the signals in all this noise, "deep learning" is an important tool in the data scientist's machine-learning repertoire. As van Rijmenam discusses, deep learning, which often leverages neural networks, helps to extract sense from streams that may involve a hierarchical arrangement of semantic relationships among component objects. "[Deep learning] is capable of breaking down different characteristic constituents in the data and uses those characteristics to learn itself different combinations of those characteristics to know what it sees (a face for example) or what to do (walk for example with robots)."
Clearly, machine learning is a fundamental tool in building a world that can sense and react to dynamic, distributed patterns. Humanity's ability to detect and respond to real-time threats and other issues -- terrorist activities, natural disasters, hurricanes, and so forth -- depends on automated sifting, sorting, and correlation of events across myriad streams.
Without this ubiquitous capability, the human race risks drowning in its own big data.