Banner Image

Next Gen CMI

Learning is set to evolve: Beyond the Blackboard

 
May 2, 2016

The process of imparting education has undergone a sea change. With the passage of time, the Nalandas and Taxilas of the East, and the cathedral and monastic schools of the West have given way to a different breed of universities and schools. And with the maturing of technology, education as we know it now, is set to evolve further.

Educational institutions are moving from the traditional methods of teaching to custom methodologies that promote ’personalized and adaptive’ learning. The widespread use of technology in education is producing enormous amounts of data. This is where data mining and machine learning techniques can come into play to help institutions to ‘teach by learning’. While we may still be seeing only the tip of the iceberg, here are a few futuristic applications that come to mind:

The right guidance: One of the biggest challenges the education industry faces is identifying areas of improvement and providing recommendations on learning objects well in time. Learning Management Systems (LMS) provide the perfect answer here. An LMS can help educational institutions capture information on a student’s usage and preferences, such as the type of learning object or training accessed, the frequency and time of access, as well as the duration of time spent on a particular course. Since tests are often linked to learning objects, these systems can make it a lot easier for teachers to ascertain a student’s understanding of a particular subject area. Through Bayesian analysis, the probability of students succeeding in summative assessments based on their performance in formative assessments can be inferred. Teachers can use this information to identify areas of improvement and help students stay on course for their learning objectives.

Simplifying evaluations with automated grading of constructed response: Educational institutes can now use machine learning techniques to automate the grading of constructed responses such as short answer questions or essays. Using word count and frequency of usage, topics discussed, and other such aspects as evaluation criteria, these automated systems will not only shorten the grading cycle time, but also allow near accurate prediction of final scores with the help of regression analysis.

Custom curriculum: Revising curricula and pedagogy has always been a time consuming and expensive exercise, the success of its outcome often unpredictable. Machine learning based prediction can help manage the content and mode of teaching. Based on the performance and behavior of students – for instance, if they drop a course or switch a major subject – program designers can redesign the curricula.

A personal tutor: Machine learning plays a significant role in cognitive psychology. These days, cognitive models are being developed to simulate the problem-solving ability of a student. This includes understanding the mechanism of human thought process, language processing, mathematical reasoning, learning, and memory. An Intelligent Tutoring System (ITS) is a commonly used cognitive model for learning. By using adaptive learning techniques built on a well-defined model of knowledge, it can provide focused guidance to students. It analyzes the strengths and weaknesses of each student related to problem solving, and provides individualized instructions.

Detecting misuse: Intelligent tutor systems often pose the ‘next best’ question to students. A few such iterative cycles result in students knowing the correct answers to questions, and advancing in the tutoring system, without really having the required knowledge. Rather than learning, this allows the students to cheat the system. The ‘training’, hence, proves to be ineffective and may even lead to drop-outs. Analyzing a student’s responses using machine learning techniques can help identify gaming patterns and aids in redesigning tutorials.

These changes in the education industry may take a few years to gain ground, but they come with the promise of reducing the manual effort at the teachers’ end. And this is good news for everyone, since these techniques will allow teachers to use the time saved to guide and support active learning. Put together, this will also result in synergies between education, cognitive science, and machine learning. The scores are out on this one – machine learning techniques are set to change the learning process, and the broader education system, for the better.

Anjan Dutta is an Enterprise Architect with the Media and Information Services business unit at Tata Consultancy Services (TCS). He has around 20 years of experience spanning areas like publishing technologies, smart content management and replication, enterprise search, natural language processing, and machine learning. He helps create solutions for leading K-12 education and scholarly publishers, financial and retail information service providers, and so on.