5 Technologies to learn for 2021 and Beyond
29 Sep 2020 | Nikhil Nair
Emerging tech has changed the world in unprecedented ways in the last decade, making the world almost movie-like in its advancements. Space travel has come to the private sector, and someone somewhere conceived of driverless cars – possibly as a reaction to seeing several vehicular accidents. Who knows.
We’re just past the midpoint of 2019, and it is a good time to take stock of the trends that we can expect to transform our future in 2020. And perhaps gain an edge by upskilling early in those areas?
The next level up from natural language processing (NLP) and machine learning is the combination of the two: augmented analysis. It was coined by Gartner as the “future of data analytics”. Let’s take a closer look at what exactly this new term actually means.
As with everything else in data science, one starts with data. It is gathered from a multitude of sources, both public and private, then automatically processed to yield insights that are useful to business operations and readable by human beings.
The key difference between regular analytics and augmented analytics is that a data scientist isn’t required to interpret patterns in the data. It is understandable by the non-tech members of the team.
This isn’t to say that organizations will be able to do away with data scientists altogether; far from it. Augmented analytics will help with the drudgery associated with analysis, like the collection of data and preparation, thus assisting data scientists in other, more intelligent aspects of their work.
The next technology on our list strays into the domain of physics, at least to grasp the underlying concept: superposition. In classical computers – yes, in this context, normal computers are considered ‘classical’ – the fundamental base unit of all operations is a bit. A bit can take one of two states: 1 or 0. Never both at the same time, and thus it is considered to be binary in nature.
Quantum computing uses something called a ‘qubit’ which, contrary to an ordinary bit, can hold both 1 and 0 at the same time. This phenomenon is known as superposition.
Superposition isn’t a magical state, but more the ability to predict the outcome of a state before it actually occurs. If this is done for lots of qubits, their proposed outcomes can be mathematically related to each other, another concept known as entanglement. Thus the outcomes are calculated before they ever occur.
The reason that quantum computing is so next-gen is that it is super fast. However, there are still kinks in the system, leading to unreliable outcomes. So their widespread use is still debatable at this stage.
In spite of deep learning is as amazing as it is, it still has limitations. It excels at relating inputs to outputs when both are clearly defined. It relies on existing patterns derived from processing large datasets to make decisions. There is no inherent reasoning, like a human being can perform.
For instance, deep learning can learn to detect a disease rapidly based on the analysis of symptoms, checked against a library of saved symptom patterns, using decision trees to arrive at that conclusion. However, it cannot understand what that disease is.
Deep reasoning aims to mimic human intelligence, by using common sense.
The concept of Small Data isn’t a new one. Like its counterpart Big Data, it has been around for a long time, but it is experiencing a resurgence in popular consciousness.
We all know that Big Data refers to large volumes of data that are analyzed to extract patterns through computations, which then, in turn, yield actionable business insights. The quantity of data makes it impossible for an individual or even a team of individuals to parse manually, and that’s where small data picks up the slack.
But first, a definition by Allen Bonde, research director at Forrester: “Small data connects people with timely, meaningful insights (derived from big data and/or “local” sources), organized and packaged – often visually – to be accessible, understandable, and actionable for everyday tasks.”
In short, small data can be used to draw intuitive conclusions from data that can be actioned faster.
By no means is small data meant to replace big data; it is more to provide actionable insights faster, while big data is being processed.
Unsurprisingly, as the name suggests, computer vision is about teaching computers how to see. Simple enough, conceptually, it is incredibly hard to bring into being.
There is often confusion between digital image processing and computer vision. However, the former is quite simply a computer being able to identify components of images, whereas the latter is a computer being able to identify objects in images, understand what those objects are, and most importantly what to do about that information.