Inside The TUKL Lab In NUST-SEECS – How Far Are We In The Machine Learning Domain?

Machine Learning is the talk of the town these days. Conventional processes which used to be digitally transformed by IT solutions are now even more fast-tracked thanks to the advancement in Artificial Intelligence. We talk to Prof. Dr. Faisal Shafait from NUST-SEECS to find out what research is currently being done at the campus and how the AI/ML landscape of Pakistan currently looks like.

With extensive research experience and knowledge in the domain of machine learning and pattern recognition, Dr. Faisal Shafait is highly regarded as a teacher and researcher at NUST. In fact, it is his brilliant contributions to document image analysis and computational forensics that have seen him secure the IAPR award and make history. Dr. Faisal is the Director of TUKL R&D Center at NUST-SEECS which conducts research in AI/ML and linked domains.

Q1. Can you tell us about the lab, how many people are involved and the main thematic areas of research being focused on in the lab?

TUKL R&D Center strives to carry out applied research to bring international cutting-edge, technological advancements to Pakistan for solving local problems and to nurture young talent to compete in the international market. It has become a leading hub for collaborative research & development in Machine Learning & Artificial Intelligence, driven by values of Innovation, and Integrity with primary focus on cultivation of talent & technology. Here at the lab, we make scientific innovation that can be scaled to help solve local and global problems. The main thematic areas of research are 

1. AI-based Environmental Monitoring where efficient algorithms are being developed to monitor deforestation, water resource estimation, crops yield estimation, and water quality assessment through satellite imagery. 
2. Artificial Intelligence in Healthcare where scalable solutions are developed for the screening procedures of endemics like Dengue Fever. The on-going pandemic of COVID-19 has also instigated our researchers to carry out intensive AI-powered algorithm development to monitor the spread pattern of the disease and relevant precautionary measures. 
3. Chatbot Development for the e-commerce industry where users can get a personalized shopping experience with an AI-powered recommendation engine.
4. Judicial Automation System for the automation of the key parts of the justice disbursement process in Pakistan through the use of Artificial Intelligence (AI). 
5. TV Analytics that comprise the development of a web-based brand coverage analytics system to monitor advertisements in multiple TV channels and companies can analyze their sponsored product coverage on a single click.

Q2. What are the initiatives being taken by the TUKL Lab which are helping or may help the Government of Pakistan?

1. In the recent pandemic times, our AI experts in the lab are conducting rigorous research for devising Artificial Intelligence-powered Machine Learning algorithms and simulation models for the forecast of COVID-19 that can be integrated with the governmental platforms to disseminate accurate information and keep track of the spread pattern of the disease. 
2. The judicial automation system project at the lab will introduce a new paradigm to the judiciary system of Pakistan where an efficient Case Based Similarity mechanism is being developed to search for relevant cases in the knowledge base for any given petition. Furthermore, a legal summarization of the individual cases will also be carried out to be presented to the honorable judge. 

Q3. How many international collaborations do you currently have?

Having raised funding of over 105 Million PKR, the Director – Prof. Dr. Faisal Shafait strives hard to maintain highly active research collaborations with various funding agencies including Higher Education Commission (HEC) Pakistan, German Academic Exchange Service (DAAD), Pakistan Science Foundation (PSF) and National ICT RDF. With immense contributions in the fields of pattern recognition and document analysis, Prof. Dr. Faisal Shafait has been awarded with IAPR/ICDAR Young Investigator Award by the International Association of Pattern Recognition which is awarded every couple of years to the most outstanding young scientist (below 40 years of age) worldwide. 

Q4. Many critics point out that research is often lost in research papers and never reaches the implementation part, how can an average researcher go about to implement his or her work and make a difference?

A research Idea is a well-sustained learning ladder that can embark upon all research projects. For a researcher to be successful, continuous discovery of the perfect improvement plan for providing a solution can be a game changer. Main crux of research is implementation. Though it is critical and obtaining the intended outcomes for is the main goal of the researchers. Strong research requires implementation and it strengthens the projects in several ways. A researcher must know how to put his ideas into the canvas to bring a better picture to the world. Here at TUKL R&D Center, we focus not only on the development of solutions but also carry out the required work for its scalability and effectiveness in the real-world. 

Q5. How can an average teenager who is currently in high school, start off in the AI and ML domain? Are there any opportunities or learning initiatives provided by TUKL for high school students?

High school students learn automation in many ways. Today multiple science, computer science-based projects are based on robotics and machine learning. Today, the students are expected to become problem solvers and leaders of the future. Education solutions for schools focus mostly on STEAM based learning that inculcates Robotics so that such students can be the entrepreneurs and solution providers in the future. TUKL lab provides the projects that can help the students of high school in many ways. Our basic yet innovative projects provide hands-on training to kids in the domains of Science, AI, Technology and Engineering. 

Q6. Does Pakistan have companies that are doing research work on the same lines as Neuralink or OpenAI?

Yes, with the advent of AI, CNN etc many startups are evolving for providing data-engineering solutions. Multiple teams excel in repetitive digital work across Data. Today research and automation is made easily accessible and in an organized manner by all solution providers. Data sourcing, data Annotation, industry classification & intellectual mapping of the data to intelligently evolve it based on ML based models is the ultimate goal of most companies today. Pakistan is expected to grow more in the coming years with more innovative startup based solution providers.   

Q7. What initiatives are your students and researchers taking to automate certain sectors of Pakistan? (i.e. record managements for vehicles, land and so forth)

ML and AI are for everyone, everyone is talking about it. Only a few actually know what to do. Today we want machines to learn and students are adapting this kind of learning. Many exhibitions are held and conferences are organized all over the country, where ML based projects are featured. Students come forward with multiple ideas that are unique some examples include Keystroke based Biometric Verification deals with continuously monitoring the identity of a user through biometric means without any extra hardware and Vehicle Recognition can detect the vehicles that are not registered to help in maintaining the security protocols. Such projects are of great help and provide solutions for multiple societies. 

 Q8. Do you agree with Elon Musk’s fears regarding AI being more dangerous than Nukes? If yes, how can we regulate or control AI from getting out of hand?

Based on his knowledge of machine intelligence and its developments, AI’s development will outpace our abilities but the reduction in human error, AI availability 24×7 and helping in multiple jobs and this digital assistance is inevitable. AI has become a part and parcel of human life, they help us make faster yet better decisions. I don’t believe that AI can be dangerous but artificial intelligence can dramatically improve the efficiency of our everything from workplaces to domestic needs. Being replaced by the machines will never be a threat to mankind, because it is the human brains that lie in creating such distinct innovations.

Q9. Finally, for a fresh graduate who has just entered the market, what would your advice be if they intend to pursue AI as a career?

To all fresh graduates out there who want to pursue their career in ML/AI, I would like to advise that the role of a machine learning engineer is at the heart of AI projects. Career-wise it has a lot of growth chances, such engineers can embark on the journey to success with AI projects and work on the frameworks of their choice to kick-start their career or shift it into high gear. It is suitable for them to have a strong background in applied research and data science. They should have the calibre to think out of the box and should be able to apply predictive models and leverage natural language processing when working with enormous datasets.