In this presentation, Romeo Kienzler discusses his autoencoder application used on IoT data. Following Romeo’s talk is an advanced presentation on neural networks by Dr. Bojan Ploj.

In this video:

Romeo begins by asking rhetorically, Why is IBM into IoT (Now)? The answer is in the numbers: there were 15 billion connected devices in 2015, with 20 billion expected by 2020. One of IBM’s contributions to IoT has been the open-source MQTT protocol, now viewed as the standard protocol for the IoT.

Romeo explains that there are two kinds of machine learning: machine learning that is (1) on historic data and (2) on online data. He contrasts and compares the two. Next, he provides a basic introduction to neural networks, in which he displays some examples of their use in such unlikely fields as visual arts and literature.

Romeo next considers some of the problems attending neural networks. For one, they are computationally complex. An answer to that is IBM’s “TrueNorth,” an advanced chip that Romeo describes in detail.

Romeo finishes with a detailed discussion of his “deep autoencoder” application.

Following Romeo’s presentation on the video is an advanced presentation on neural networks by Dr. Bojan Ploj entitled, “It’s time for a change: Two new deep learning algorithms beyond backpropagation.”

Resources

Discovering Data Science with Romeo Kienzler

Follow Romeo as he tackles the most difficult challenges in data science.

Join The Discussion

Your email address will not be published. Required fields are marked *