Machine Learning and the Data Revolution
If you keep yourself abreast with the news, there is a high chance that you must have heard the term – ‘Machine Learning (ML)’ in recent times. Going by the meaning, ML might either seem like a herculean task to make machines learn or some sort of sorcery whereby machines acquire the requisite intelligence. But in reality, machine learning is neither a technical jargon nor a superstition. It simply uses the different concepts from computer science, data management, electronics and manufacturing technologies to allow machines to carry out their tasks without or with little help from humans. We will try to understand more about MLand how it is driving the latest technological revolution.
Machine Learning – What is it?
As already mentioned, ML is a subset of Artificial Intelligence. AI is the field that focuses on imparting intelligence to machines so they can perform the prescribed tasks with little human intervention. ML is an important component of AI, because it actually deals with the methods and techniques used to make a machine ‘intelligent’. In other words, ML is a discipline dedicated to teaching machines how to learn from their surroundings. These learnings are then applied by the machine either to make decisions or predictions. Thus, machine learning is a technology that empowers machines to not only continuously learn from their environment but also provide them with autonomous decision-making abilities.
Thus, contrary to popular belief, machine learning methods do not cover the entire spectrum of human intelligence but focus only on a narrow domain demanded by automation.
How does ML work?
The cognitive function of learning almost exclusively belongs to human beings. It is a highly complex and sophisticated process of the human brain. I say this because, the depth and expanse of knowledge acquisition that has been attained by humans is hitherto unmatched by other living beings. Hence, you might wonder, how do machines that lack consciousness and a brain learn? The answer lies in mimicking the brain’s process in reaching a conclusion or making a decision. Let us understand this further.
Imagine you are in an ice cream shop with different flavour options to choose from. In such a scenario, how do you decide which ice cream to have? Let’s unravel your thought process behind making a choice. There are a few different cases in this example –
1) You have a favourite flavour that you always order.
2) Out of the many different options available, you’ve narrowed down to 3 or 4 flavours based on your past experiences and your knowledge regarding the taste and ingredients of various flavours.
3) You are completely clueless as you’ve never had any of the available ice cream flavours, nor does the shop provide information regarding ingredients and tastes.
In the first case, the decision is based on your previous experience wherein you had encountered a similar situation. The second case presents before you a new situation but you are equipped with enough data to make a decision, that is based on your preferences and past experiences. But the third situation proves to be challenging due to the dearth of data. There is no point of reference available and you would purely have to decide using your instincts.
Thus, in the above example, the significance of past experiences and data availability is evident for making a decision. This is how our brain also reaches conclusions through analysis of past and present datasets. This is the exact process employed by machine learning models.
The Feedback Loop, Machine Learning Models and Data Mining
A machine learning model is software that has been trained for using data to recognise patterns within that dataset. In order to improve the accuracy of the ML model, it is trained using a certain set of data known as the ‘training dataset’. This model is then tested using a ‘testing dataset’ so as to calculate the model’s accuracy. This is done by juxtaposing the outcome of the ML model with that of the available results within the testing dataset. The more data used in training the model, the more accurate it becomes in predicting values or deciding outcomes.
From the aforementioned instance, you can ascertain the importance of data in training a machine. Hence, providing adequate and relevant data is super important in ensuring that an ML model works accurately. This is why data mining or the process of extracting data and identifying patterns within it is an indispensable operation within the ML domain. Thus, data analysis by a ML model is possible first due to data mining and then training the model based on the dataset. Apart from data analysis, ML also involves data preprocessing, data management and data visualisation.
Training and testing of a model does not cease after one iteration. It is a continuous process that enables the model and thereby the machine to learn about newer circumstances and equip itself to face them. This is known as the ‘feedback loop’ wherein the algorithm (model) collects data from its ambient conditions and uses the learnings to improve its efficacy and efficiency. Without the ability of the feedback, there is a high chance of the model getting redundant over time, as the surroundings and datasets evolve from the ones that were used to train it. Hence, feedback loop is a highly coveted characteristic of any machine learning model.
Applications of Machine Learning Algorithms –
Applications of ML models are vast and encompass almost all fields of modern economy. Right from predicting the COVID-19 infections curve to suggesting videos on YouTube, ML has become ubiquitous. With the proliferation of robotic and Internet of Things (IoT) devices, machine learning models are seeing an increase in demand. They are also used in correctly predicting the path of a hurricane, to estimate food production within a geography and also in the differentiation of benign and malignant tumours. While doing so, the model takes into account the past data, current trajectory and data peculiarities, if any, in order to provide results. Again, the optimum amount of accurate and complete data ensures the best possible results from the model.
Any application that enables generation of quantifiable data can be turned into a machine learning model to achieve improved accuracy and high operational speed over that of predictions by humans.
Economic Impacts of Machine Learning –
The positive role of ML and artificial intelligence in the research and development (R&D) sector of the global economy is significant, given the reduction in time required to dole out new products in the market. These technologies are hailed as being disruptive when it comes to wage and income distribution as well as economic and social inequality. Though many low skilled jobs shall be replaced by machines, increased demand for highly skilled workers is estimated to drive up wages and thus lower the economic inequalities. Evidently, ML and AI are driving economic growth by augmenting the finite labour supply. In the era of globalisation, the labour market is no longer constrained by geography as the gig economy has taken off due to adoption of latest technological standards. Hence it is now possible to earn in USD while spending in INR. This ‘freelancing economy’ is estimated to grow further as the AI and ML technologies proliferate.
Thus, ML is a technology that has revolutionised the fields of data collection and analysis to bring in newer methods and tools for business analytics, in order to drive growth and prosperity. It is said that one should always swim ‘with’ the current rather than ‘against’ it. Looking at the current technological as well as job market trends, ML certainly seems promising. This is owing to the increased dependence of businesses on data collection and its analysis in order to make decisions. The novel professions of data scientists, data analysts, business analysts, etc. have emerged as a result of predominance of data in automation and decision making processes. Thus, the field of machine learning has turned out to be a truly revolutionary technology that promises to drive growth as well as equality.
Vishvali Deo is an E&TC (Electronics and Telecommunication) Engineer by education and Software Engineer by Profession. She believes that 'Technology is a Great Democratising and Equalising Force' and hence is on a mission to make the general public understand seemingly complex technologies in a simple manner.
She is convinced that the root of today's world problems lie in the past, hence she has also pursued post-graduation in History. She has a keen interest and a good grip over Economics, Political Science and Environmental Engineering. She has a penchant for working with Women and spreading Digital Literacy amongst them, with the aim of their empowerment. She also strives to provide Free Quality Education to children and counsels young adults. Besides, she is also skilled at Public Speaking, having won many awards in Elocution & Debate Competitions and Technical Paper Presentations.