DAVID TOLFREE, SCIENCE AND TECHNOLOGY WRITER
We live in an increasingly connected world where data is continuously created and transmitted. Advances in communication and computational technologies have produced unprecedented levels of sophistication in data collection, analysis and storage methodologies. It is estimated that there are currently 4.57 billion internet users in the world, growing at an annual rate of more than 8 percent1. Our millions of computers, mobile phones and other interconnected devices all form part of the Internet of Things (IoT), so are major generators of data.
The global market intelligence company International Data Corporation (IDC) recently published a white paper titled Data age 2025, the digitization of the world from edge to core, which predicts that worldwide data creation will grow from its estimate volume of 33 zettabytes in 2018 to 175 zettabytes by 2025 (a zettabyte equals one sextillion (1021) bytes) 2. The paper provides insight into the challenges facing the world in the handling of data and the information it relays. An area of growing concern is the accurate interpretation of data, since it is used extensively in informed decision-making.
The coronavirus pandemic has highlighted the issues surrounding collecting and analysing reliable data, and the serious consequences that can arise if said data is misinterpreted. In the case of the latter, political bias and the human capacity for scepticism and self-deception distorts information. Statistical tools such as charts, graphs and tables are often used to display qualitative and quantitative data to help in the understanding of results, but these are only effective if interpreted correctly.
The interpretation and correct use of data is now of the utmost importance in every sector of industry and public services, particularly as artificial intelligence (AI), machine learning and automated control systems play an increasingly dominant role in society.
Artificial intelligence (AI) and big data
The realm of big data is being embraced by companies globally. Big data is inextricably linked to AI and has thus accelerated its adoption in many sectors.
There are various definitions of big data, but the Oxford Dictionary one is: “Extremely large data sets that may be analysed computationally to reveal patterns, trends, and associations, especially relating to human behaviour and interaction.” These data sets can be in the terabytes to zettabytes range, meaning their handling is often beyond the capability of most commonly used software tools.
AI uses algorithms to enable computers to analyse large data sets. Algorithms are sets of instructions in computer programmes designed to reduce the time that humans require to analyse such data, but problems arise if the results contradict expected outcomes. UK examination boards used an algorithm to predict students’ marks because of the coronavirus pandemic. However, this algorithm contained some criteria that related to a school’s status and past exam results, and misinterpretation of some of its instructions resulted in a downgrading of the marks predicted by teachers and achieved during mock exams. After a great deal of protesting by students and their parents, the examination boards agreed to reject the marks predicted by the algorithm in favour of those predicted by teachers and achieved during mock exams.
For many years, AI, algorithms and supercomputers have been used to predict regional weather patterns based on the interpretation of past data. This enables more accurate models to be made for weather forecasting. Masses of data is collected daily from over a thousand weather satellites circulating the Earth, in addition to a greater number of ground-based weather stations. Businesses, government departments and regional public bodies use algorithms extensively to predict outcomes and thus assist in decision making. Also, Amazon, Facebook, Google and other online multinational companies are reliant on them for their continued success.
The subject of AI and big data will be covered in depth at the virtual conference COMS World 2020 on October 19–21, details of which can be found at www.comsworld2020.com.
Interpreting research data
Data collection, analysis and interpretation are key elements in all areas of research. Most of my time in nuclear research had to be spent interpreting experiment results, and the data often produced unexpected outcomes. This meant that more data needed to be collected in order to understand the results. For research scientists, this iterative process is normal to verify discoveries.
Similar processes are used to interpret results from human trials of drugs and vaccines. New drugs and viruses produce many different reactions in humans so collecting and understanding such data is vital in testing the efficacy of vaccines or drugs.
Data interpretation from images
There are numerous applications for image analysis, including autonomous driving, medical imaging, smart video surveillance for security, visual inspection for process control in manufacturing and astronomical observations. Indeed, advances in nanoresolution display technologies have meant that the image recognition market is expected to grow from US$15.95 billion in 2016 to $38.92 billion by 2021 at a compound annual growth rate (CAGR) of 19.5 percent3.
Image recognition and processing forms an essential part of an autonomous (self-drive) vehicle (AV). An autonomous car receives real-time sensory input from several sensors, including camera, radar and light detection and ranging (LiDAR), to form digital maps. The image data from these maps is then interpreted by an AI software-driven computer to control the car.
Crew Dragon, a space vehicle built by Elon Musk’s company SpaceX, serves as an excellent example of the levels of accuracy and precision achievable thanks to touchscreen automated control technologies4. This space vehicle is designed to fly autonomously for the duration of missions, including docking at the International Space Station (ISS). Imaging sensors such as cameras and LiDAR on the nosecone of the space vehicle are used to guide its approach to the space station. These feed data back to the flight computer, informing it, for example, of the space vehicle’s distance from and relative velocity to the space station. The flight computer then creates algorithms based on this data that determine how the space vehicle’s thrusters should be fired for it to get most effectively to the docking target.
The importance of accurate data interpretation cannot be underestimated, since it underpins all automated control systems, systems that society is growing increasingly dependent on and are changing our lives.
References
1Digital around the world (web page). DataReportal. Available at: https://bit.ly/2QRMfo0
2Reinsel, D., Gantz, J. and Rydning, J. (2018). Data age 2025, the digitization of the world from edge to core [white paper].
Available at: https://bit.ly/31TXKBB
3Image recognition market by technology (digital image processing), component (hardware, software, service), application (augmented reality), deployment type (on-premises, cloud), industry vertical and region-global forecast to 2021 [report]. MarketsandMarkets. Available at: https://bit.ly/3gXHDag
4Rincon, P. (2020). What is the SpaceX Crew Dragon? [press release]. July 31. BBC News.
Available at: https://bbc.in/32ZFUN4