The emergence of the Internet and its worldwide spread caused a sharp increase in the amount of digital information being generated, stored, and made available for analysis. The digital universe doubles roughly every two years, promising to reach a record of 40,000 exabytes, or 40 trillion gigabytes, by 2020, containing masses of raw data that might be valuable if analyzed. The human brain physically cannot process such a flood of information – but machines can.
Intelligent machines that work and react like humans, though much faster and with fewer mistakes, seemed to be in the realm of futuristic sci-fi movies just a decade ago. Today, however, artificial intelligence is clearly moving into the mainstream, and behind the vague buzzwords, there is a set of pretty concrete technologies and techniques – for instance, computer vision, natural language processing, robotics, and machine learning.
Being a key part of AI, machine learning empowers computer algorithms to explore the digital universe for analytic value – in other words, to use huge chunks of data to identify hidden patterns, recognize trends, accurately predict future outcomes, and make data-driven decisions. The domains adopting machine learning range from government, manufacturing, energy, and telecommunications, to healthcare, retail, banking, and finance.
Though ambitious ideas and deep-seated cynicism connected with machine learning tussle with each other in the marketplace, experts forecast that the era of smart machines will become one of the most disruptive phases in the IT history. The majority of researchers expect the machine learning market revenue to exceed $12 billion by 2020, up from 2.5 billion in 2014, and Gartner predicts that one fifth of enterprises will have positions devoted to “monitoring and guiding” machine learning within a couple of years.
Interestingly, BCC Research divides the global market for smart machines into five segments: neurocomputers, expert systems (e.g., medical decision support), autonomous robots (including self-driving vehicles), smart embedded systems, and intelligent assistance systems. Expert systems are currently the largest segment, but the autonomous robot category is projected to dominate the overall market by 2024.
Auriga Delivers Machine Learning Projects
Thus, as machine learning is gaining wider popularity in multiple industries, more and more companies realize the benefits and opportunities it offers and want to apply it to their businesses. In recent years, Auriga’s engineers have delivered a number of exciting machine-learning related projects for major customers from various sectors. Here are only three examples of how machine learning is useful in the real world.
For healthcare, Auriga has created a high-load cardiac monitoring system based on a three-layer neural network trained to recognize critical cardiac conditions. An in-depth statistical analysis of a massive amount of cardiac data coming from the sensors of portable ECG Holter devices, combined with weather data, enables continuous monitoring of hundreds of thousands of patients simultaneously, early detection of emergency conditions, and even prevention of cardiac disorders. All this makes the system comparable to real-time treatment.
As part of a big driving automation project, Auriga has developed a semi-automated object identification tool to label video, LIDAR data, and speed data, to create a stream that can be used for machine learning. With some manual operator assistance, the application uses the uploaded data to automatically recognize and label objects with a predefined set of labels, including lane markings, traffic lights and signs, trees, other vehicles, cyclists, and pedestrians. Importantly, video of driving in poor weather conditions forms a considerable part of the created data lake. The labeled data forms a data stream for subsequent machine learning and driving automation.
More recently, Auriga has also participated in a big smart agriculture project by developing a drone software utility to monitor the condition of trees in an orchard. Our engineers deployed a ready-made neural network that was already trained to recognize a large number of various objects and continued training it on the orchard data collected by a camera-equipped agricultural drone. The solution allows farmers to track their orchards remotely, easily assess the trees’ health, fine-tune irrigation, schedule resource allocation, and hence optimize performance.
As Elena Baranova, Engineering Director at Auriga, noticed,
We live in a time when the power of high-value prediction and data-driven decision making gives enterprises a strong edge over competitors. If you want to make your business smarter and more efficient, machine learning is often a key.
