While machine learning is a relatively new concept for many, there are an increasing number of platforms and services that are adopting this construction of algorithms. ROBERT SCOTT, SafariNow CCO, talks about the relevance of supervised and unsupervised machine learning.
Given the massive increase in the amount of data that companies – of all sizes and across sectors – are generating, it is no surprise that data analytics and machine learning are fast becoming key components of every innovative company’s toolkit. For the uninitiated, machine learning refers to the way in which companies can now leverage computing power to find important patterns within their data – and then use these patterns to improve their service or product offering.
Because of the sheer volume and complexity of the data being created today, it is often far beyond the capacity of any human – no matter how analytically gifted – to find any relevant trends or insights within what has been tagged ‘Big Data’.
Notably, one of the big differences between machine learning and computer-assisted analysis (where humans are involved) is that the recent breakthroughs in machine learning enable computers to teach themselves how to solve problems. So previously, when humans were directing computers, they were limited to very direct questions and answers (for example, “what is my top selling item?”) and required the person using the machine to dictate which method to use to the solve the problem. Now, machine learning enables computers to find answers in ways that are unguided by human intervention.
Although it is a relatively new and novel concept for many, the technology has already been applied to platforms and services that we use daily. Take Google Search, for example. When we enter a search term, Google uses elements of machine learning to analyse our behaviour once the first results have been served up (i.e. did we need to type in the same search term again, or did we follow some of the top links provided?) and then refines and improves its service according to the data. Other examples include Google’s self-driving car, how Netflix suggests which movies you should try next, and how a dating site suggests which people are most likely to be a suitable match for you…
As with most technological tools today, almost any company or sector can leverage machine learning to better serve their customers. The challenge for companies is to recognise where – and how – certain insights and trends can improve their product or service offering.
Within the travel sector, we have identified various areas in which machine learning can be applied in order to fine tune our offering and help travelers locate their dream destinations. One of the great benefits of this tool is that it often finds relationships between factors that are completely unexpected and unplanned.
Machine learning has led us to the insight, for example, that some accommodation providers have a preference for prioritising requests from customers who would like to stay with them in the next few days – whereas other providers would much rather prioritise requests far in advance (for the school holidays, for example). Often, it is these unexpected – or unplanned – insights that can be the most beneficial for customers.
As an online travel aggregator, there are in fact infinite possible use cases for machine learning – and we are at the tip of the iceberg in terms of harnessing its potential to improve our offering to consumers looking for the next adventure.
Looking ahead, machine learning will perhaps become a standard application within the travel and e-commerce environment. Companies that are open to innovative ways of finding insights in their data can ultimately serve their customers more efficiently – and even develop closer relationships with them in the long-term. The key for companies is to keep an open mind as to whether or not their long-held beliefs about what customers want is actually supported by the data.
By always remaining alert to new patterns and insights, companies can make adjustments – both big and small – to enhance their offering.
Huawei Mate 20 unveils ‘higher intelligence’
The new Mate 20 series, launching in South Africa today, includes a 7.2″ handset, and promises improved AI.
Huawei Consumer Business Group today launches the Huawei Mate 20 Series in South Africa.
The phones are powered by Huawei’s densest and highest performing system on chip (SoC) to date, the Kirin 980. Manufactured with the 7nm process, incorporating the Cortex-A76-based CPU and Mali-G76 GPU, the SoC offers improved performance and, according to Huawei, “an unprecedented smooth user experience”.
The new 40W Huawei SuperCharge, 15W Huawei Wireless Quick Charge, and large batteries work in tandem to provide users with improved battery life. A Matrix Camera System includes a Leica Ultra Wide Angle Lens that lets users see both wider and closer, with a new macro distance capability. The camera system adopts a Four-Point Design that gives the device a distinct visual identity.
The Mate 20 Series is available in 6.53-inch, 6.39-inch and 7.2-inch sizes, across four devices: Huawei Mate 20, Mate 20 Pro, Mate 20 X and Porsche Design Huawei Mate 20 RS. They ship with the customisable Android P-based EMUI 9 operating system.
“Smartphones are an important entrance to the digital world,” said Richard Yu, CEO of Huawei Consumer BG, at the global launch in London last week. “The Huawei Mate 20 Series is designed to be the best ‘mate’ of consumers, accompanying and empowering them to enjoy a richer, more fulfilled life with their higher intelligence, unparalleled battery lives and powerful camera performance.”
The SoC fits 6.9 billion transistors within a die the size of a fingernail. Compared to Kirin 970, the latest chipset is equipped with a CPU that is claimed to be 75 percent more powerful, a GPU that is 46 percent more powerful and an NPU (neural processing unit) that is 226 percent more powerful. The efficiency of the components has also been elevated: the CPU is claimed to be 58 percent more efficient, the GPU 178 percent more efficient, and the NPU 182 percent more efficient. The Kirin 980 is the world’s first commercial SoC to use the Cortex-A76-based cores.
Huawei has designed a three-tier architecture that consists of two ultra-large cores, two large cores and four small cores. This allows the CPU to allocate the optimal amount of resources to heavy, medium and light tasks for greater efficiency, improving the performance of the SoC while enhancing battery life. The Kirin 980 is also the industry’s first SoC to be equipped with Dual-NPU, giving it higher On-Device AI processing capability to support AI applications.
Read more about the Mate 20 Pro’s connectivity, battery and camera on the next page.
How Quantum computing will change … everything?
Research labs, government agencies (NASA) and tech giants like Microsoft, IBM and Google are all focused on developing quantum theories first put forward in the 1970s. What’s more, a growing start-up quantum computing ecosystem is attracting hundreds of millions of investor dollars. Given this scenario, Forrester believes it is time for IT leaders to pay attention.
“We expect CIOs in life sciences, energy, defence, and manufacturing to see a deluge of hype from vendors and the media in the coming months,” says Forrester’s Brian Hopkins, VP, principal analyst serving CIOs and lead author of a report: A First Look at Quantum Computing. “Financial services, supply-chain, and healthcare firms will feel some of this as well. We see a market emerging, media interest on the rise, and client interest trickling in. It’s time for CIOs to take notice.”
The Forrester report gives some practical applications for quantum computing which helps contextualise its potential:
- Security could massively benefit from quantum computing. Factoring very large integers could break RSA-encrypted data, but could also be used to protect systems against malicious attempts.
- Supply chain managers could use quantum computing to gather and act on price information using minute-by-minute fluctuations in supply and demand
- Robotics engineers could determine the best parameters to use in deep-learning models that recognise and react to objects in computer vision
- Quantum computing could be used to discover revolutionary new molecules making use of the petabytes of data that studies are now producing. This would significantly benefit many organisations in the material and life sciences verticals – particularly those trying to create more cost-effective electric car batteries which still depend on expensive and rare materials.
Continue reading to find out how Quantum computing differs.