AI needs curiosity and creativity
Preeti Tamrakar is a data scientist in the AI-COE area at Cognizant.
She has worked as a researcher on various classification approaches and feature selection methods. In addition, she creates AI/DL/ML based applications for various clients.
INDIAai interviewed Preeti to get her perspective on AI.
What motivated you to switch from academia to the AI industry?
I wanted to contribute more to solving real problems and grow faster. I wanted to explore both sides of the AI field, academia and industry. After my MTech from IIT (ISM) Dhanbad, I spent eight years in academia, and after PhD it was time to experience the industry. The AI industry is the fastest growing and developing market. It’s been two years in the industry, and I’m determined to use my research skills to help the industry and give back to the world.
What initial challenges did you experience during this transition? How did you overcome them?
Academia is research-oriented, but industry is execution-oriented, which means there is a difference in the working nature of academia and industry. The sector may prefer to have something other than pure research minds on the team, but is looking for people with problem solving and programming skills. I had to change my nature of work to adopt a culture of fast moving and financial benefits of the industry. I worked on presentation skills to overcome these gaps, focused on productivity for quick outcomes and multitasking. Advertising myself on a social media platform like LinkedIn was a critical turning point for me to be accepted in the industry.
Tell us about your research in data science. What were your research contributions?
During my Academia tenure, I worked on different classification and feature selection techniques during my research. The research aimed to construct an efficient lazy learning associative classifier to improve classification performance. Thus, different associative classification approaches were studied and new innovative techniques were proposed in LLAC. Lazy learning is suitable for complex and incomplete problem domains, where datasets change frequently.
Classification is one of the essential methodologies used to predict group membership for unseen data instances. It uses supervised learning, in which class label is already present. This is a two-step process. In the first step,
- a model is constructed using the training datasets.
- The model is represented as mathematical formulas, decision trees or classification rules.
- The second step is model use for the classification of unknown cases.
- The model is validated using the test dataset.
We can use this for applications where accuracy over time constraints is more important – for example:
- fraud detection,
- spam filtering,
- cancer diagnosis,
- weather forecast,
- Churn tracking etc.
Currently, my industry research work is focused on Long Short Term Memory (LSTM), which uses gates (Forget, Input, Output) to discard valid information and utilize useful information. It is used to build a smart model for the applications such as alert prediction, log anomaly detection, no code environment and time series forecasting.
What is your day-to-day role at Cognizant Technology Solutions as an Associate Projects (Data Science)?
In Cognizant I work in the department of AI-COE. My responsibilities are to develop AI/DL/ML models for different clients. Python language is used in my work because many tools are available like Scikit Learn, Tensorflow, Theano, Keras, matplotlib, numpy and pandas which help to program the concepts and algorithm. Based on this, user interface tools are developed, which enable users/clients/clients to build their software model for their application without coding.
The building model should be advertised to the customer by giving demonstrations and presentations. All models may not suit customer requirements/needs, so feedback becomes essential to improvise and make it ready or valuable for the customer.
What qualities do you look for in a startup in the AI field?
Determined to work on AI with curiosity and creativity: A solid background in math and statistics is helpful in traditional software engineering, but mandatory for machine learning. The fundamental knowledge of statistics, probability and mathematics enables machine learning engineers to understand which algorithms best address a problem and how to optimize outcomes.
Employers are looking for individuals with curiosity and creativity to excel in AI. These skilled minds are the ones best able to solve the unclear problem and bring clarity to the possibilities that exist in machine learning.
Tell us about your long-term AI research goals.
Having had insightful experience credited with IIT and VIT culture and vast expertise in data science principles and practices, I have gained sound knowledge during academia and while executing projects; a deep understanding of the data science domain. I aim to build a complex AI system with a no-coding environment for users (such as regular people or non-AI background users) to avoid tedious tasks and be productive. The building model will help people avoid mistakes and complete tasks accurately.
What advice do you have for students and professionals interested in working in artificial intelligence?
Artificial intelligence is a branch of computer science that includes the development of intelligent computer systems to solve problems that are not necessarily only related to computers, but in any domain or industry.
AI has been used in various applications including data mining, machine learning, robotics, medical diagnosis and treatment, and many others. Many of AI’s breakthrough technologies, such as “natural language processing,” “deep learning,” and “predictive analytics,” are familiar buzzwords. Thanks to the latest technologies, computer systems can understand the meaning of human language, learn from experience and make predictions.
I urge students and professionals to develop skills (coding and analytical) to find hidden problems in dumped massive data and solve them in many ways. But unfortunately, it is often recognized that problem solving and communication skills are the essential skills an employer looks for in an aspirant.
Can you recommend any AI books or research papers for people just starting out in the field?
For those planning to seek a place in the AI field, start today by preparing yourself with the tools needed to successfully perform the job. Obtaining certifications in domains like machine learning and AI is a good place to start, and with the right training, the opportunities are endless.
I suggest starting with udemy online course: “Machine Learning AZ: Python & R in Data Science” https://www.udemy.com/course/machinelearning/
and read articles about: