Introducing deep learning and long-short term memory

Detecting anomalies in IoT time-series data by using deep learning Romeo Kienzler
Published on May 16, 2017

facebooktwitterlinked Ingoogle+e-mail This page 0 content series: This content was part 1 of 5 in the Seri ES: Developing cognitive IoT solutions for anomaly detection by using deep

Although predictions are always controversial, Gartner says that there are 8.4 billion connected IoT in devices (not Counting smartphones) and some analysts say that by 2020 there would be billion. Even if over-estimated, soon the number of IoT devices would exceed the number of humans on the planet. And guess what, all of these devices are continuously generating data; Data is useless unless can analyze it.

A cognitive system provides a set of technological capabilities such as artificial intelligence (AI), natural language Processing, machine learning, and advanced machine learning to help with the analyzing all that data. Cognitive systems can learn and interact naturally with humans to gather insights from data and help for you to make better de Cisions. In my last article i stated cognitive computing??? just Human-computer (HCI), it is interaction mac Hine learning driven by powerful algorithms (models) and nearly unlimited data processing.

To understand a cognitive system that uses IoT sensors and deep-analysis, your A-learning to need the leap F ROM Advanced machine learning to neural networks. In the This article, I try to help you do that leap. In the coming weeks, I'll present three different tutorials about anomaly detection on time-series data on Apache Spark Using the deeplearning4j, APACHESYSTEMML, and TensorFlow (Tensorspark) deep learning to help you frameworks fully nd How to develop cognitive IoT solutions for anomaly detection by using deep learning. From machine learning to neural networks and deep learning

If a cognitive system is based in models, you are need to the look of the A what model is. It is a statistical model (black box), contrast to a physical model (white box) has been trained and data to learn A hidden pattern.

Look at Table 1. It contains historic data on different parameters, measured observations on a manufacturing pipeline, and a binary outcome . Table 1. Machine-learning Model

Part

No.	Max Temp. 1	Min Temp. 1	Max Vibration 1	outcome
100	35	35	12	Healthy
101	36	35	21st	Healthy
130	56	46	3412	Faulty

In this highly artificial example, the numbers speak for themselves. As you might guess, Ace temperature and high vibration LEDs to a faulty part. The root cause for this situation might is the result of a broken bearing in a machine.

In this example, a (supervised) Machine-learning algorithm are capable of considering all this data (and most) to Lear N and predict faults from pure data. The component that such a algorithm produces is called a machine-learning model.

A Special type of machine-learning algorithm is a neural network. It is highly adaptable to data, and it are able to learn any hidden mathematical function between the data and the outcome. The only catch with neural networks are the tremendous amount of computational resources and data this they need to perfor M. So why am I talking about neural networks in all? We live in a IoT world with tremendous amounts the data available and also (nearly) unlimited computational power availabl E by using the cloud. This situation makes neural networks especially the for interesting data IoT.

The

Neural networks are inspired by the human brain, and, are deep learning. The main difference between a neural network and a deep learning one is the addition of multiple neural. The most obvious example deep learning is outperforming traditional machine learning are with image recognition. Every State-of-the-art system uses a special type of deep learning neural network (called A convolution Neural networ K) to perform their tasks.

For example, deep-learning-based image recognition algorithms are capable of distinguishing healthy parts from faulty Parts in a manufacturing pipeline. I ' d call This machine Intelligence and It's available as in THE IBM Watson Visual recognition Service. For this particular example, the machine might accomplish the same task with the same accuracy as a human. The only machine advantage are that it never sleeps, never calls in sick, and never gets. And, if you are need to double the throughput, just double the amount of hardware or cloud. But applying a root-cause analysis on why parts are sometimes faulty is still the domain of human experts. However, this scenario is where cognitive solutions applying deep.

In fact, a visual recognition service returns much more information than just a binary outcome of "healthy" or "faulty." J UST like a human, the service detects structures and regions the images of that deviate. Now, if you are were to correlate all sound and vibration sensor data with all visual recognition data, I ' m sure such a system Could detect the root causes of faults as, or even better than, humans. How Artificial Neural networks work

If the IoT sensors that connected to a message broker (like the mqtt-based IBM Watson IoT Platform) are the "the", Nervou s system of cognitive solutions, then deep learning is the brain. And, to understand deep learning, you need some basic understanding of regression, perceptrons, biological and artificial Neural networks, and hidden layers. Start with linear and logistic regression

A ton of scientific literature exists on regression, so I ' ll try to give your A short-path explanation this is tailored For developers. Consider Table 2. It is the same as Table 1 except-I ' ve turned the outcome into a binary representation. Table 2. Artificial Neural network data

Part

No.	Max Temp. 1	Min Temp. 1	Max Vibration 1	outcome
100	35	35	12	1
101	46	35	21st	1
130	56	46	3412	0

It ' s pretty easy to write a piece of software to make the classification.