One of the industry’s biggest challenges in the development of the Internet of Things and the fourth industrial revolution is to provide expert data analysis, of which there is currently a pronounced shortage.
It’s an area of particular interest to Leonard Johard, founder of Calejo and researcher in artificial intelligence at Innopolis University, Russia, who believes that data is one of the most important commodities of the 21st century, and the need has outgrown the supply.
Tina Berggren Project Manager and co-founder at digital healthcare specialists Nordic Health Innovation AB caught up with Johard to find out why data analysis must learn to survive without the data in Industry 4.0.
Tina Berggren: Do we really need data analysis?
Leonard Johard: We are now in a paradigm shift and at the beginning of what’s become known as the fourth industrial revolution. Previous industrial revolutions were aimed at freeing people from the need to perform physical work. Today, the processes that previously required human intellect are increasingly automated. The unspoken goal of this is a self-organising industry that is both more efficient and more flexible than what is possible under human control. Custom artificial intelligence (AI) is now taking over industrial processes and can optimally control, monitor and adjust the processes. These systems benefit from the speed and ability to handle data volumes, which far exceeds the human capacity. The financial sector saw these benefits early on and we see that robot trading today accounts for an overwhelming majority of the financial transactions. For the production industry, the focus is on completely new production environments, with groups of machines communicating with each other and sending necessary data among themselves to optimise the manufacturing process.
Products, production, service and maintenance are also increasingly integrated. We see a future where the ability to manage the intersection between people, data and intelligent machines will be crucial for productivity, efficiency and the operation of industries around the world. The challenges are still numerous. One of the biggest is that the development of the intelligent collection and analysis of large amounts of information requires an expertise, which is currently in short supply. Modern machine learning is a complex scientific field that is often equated with an art form. The ability to attract expertise in this area is therefore crucial to the new smart industries’ business models.
TB: What is data analysis?
LJ: Data analysis is a general term for processes to extract value from the raw data. This umbrella term includes various value-adding technologies:
- Data acquisition and data processing will increase the collection of useful raw data.
- Exploratory Data Analysis (EDA) refers to studies that indicate the value of the existing data resources and appreciate the value of further processing.
- Business Intelligence (BI) includes analysis of data to support the company’s decision-making.
- Confirmatory Data Analysis (CDA) focuses on confirming or disproving existing hypotheses.
- l Predictive data analysis focuses on the creation of statistical models for predictive forecasting, classification and/or direct process control.
Predictive data analysis has seen the largest growth, at the same time as the predictive models had improved in recent years. Developments within Deep Learning have resulted in a world in which automated systems almost always outperform their human competitors, even in areas where, only 10 years ago, it was believed that the use of artificial intelligence would remain beyond the realms of possibility for the foreseeable future.Leonard Johard, founder of Calejo
TB: In what ways is Calejo a pioneer in this field?
LJ: We are now building an efficient production line of artificial intelligence designed for Industry 4.0. By systematising and automating development, we provide cost-effective access to the epitome of the data science. Our international structure and tight academic connections enables us to use both local development resources and cost-effective development remotely. Meanwhile, we are at the centre of the development, with a concentrated core of leading international researchers in artificial intelligence. This structure makes our process extremely agile and allows us to quickly adjust our models to the ever-accelerating research in this area. Our best strength is actually our contribution through research in new disruptive neural network algorithms, AI methodologies and industry perspectives.
TB: Do you think business leaders need to be more ambitious with their data strategies?
LJ: Most business leaders with their finger on the pulse have already felt the data revolution. The efficiency of extraction and refining of this valuable resource will separate failure from success throughout the next decade. Unfortunately, cloud storage is not enough. Data strategy is everything: data availability is usually taken for granted, but the data has already started to get fenced in and with a steadily increasing price tag attached. Soon the party will be over for latecomers.
Calejo data analysis already offers the sale of manufactured, artificial data for financial markets, processes and machinery.
All the data in the world will never be enough. Fortunately, we can now circumvent the data-bound limitations of traditional AI methodology, while at the same time maintaining the self-learning aspects and use select real data where and when it is available.
The potential of these artificial data have been realised thanks to the research team and a portfolio of in-house artificial intelligence algorithms.
Our core of researchers go far beyond the simplified narrative of the Deep Learning bandwagon and design algorithms from a more profound understanding of statistical learning. We continuously reverse-engineer the brain and our lab has neural network models that are a generation ahead of the competition.