Dennis Núñez-Fernández

BSc in Electronic Engineering, Universidad Nacional de Ingeniería (UNI)


How I started in research and development in AI/ML

By: Dennis Núñez-Fernández, April 2021.

My interest in the areas of AI/ML started when I was in the last years of university, there I started to become interested in topics such as Artificial Neural Networks, Signal Processing and Digital Systems. Thus, in addition to my undergraduate courses, I took courses in Control using Fuzzy Logic and Neural Networks, where I learned important fundamentals of these areas.

After finishing university, and concluding my English studies, I applied for a 6-month internship funded by UNESCO and the Government of Poland because there was an area that caught my attention: Computer Vision. I applied and had the opportunity to be accepted. A few months later, in October 2016, I traveled to Poland to start the internship, there I conducted a research project on Deep Learning for hand gesture recognition and trying to optimize computational resources by using Gabor filters. At the end of the project we were able to publish a scientific paper in a top Iberoamerican conference on Artificial Intelligence: the Iberoamerican Congress on Pattern Recognition (CIARP). That was my first scientific publication.

Returning to Peru, in 2017, I started to carry out projects on my own as an independent researcher and related to what I did in Poland. The results of these works were presented at national conferences such as INTERCON, which helped me to get more involved in research.

Later I looked for an opportunity at the Universidad Peruana Cayetano Heredia (UPCH), more precisely in the Laboratory of Bioinformatics and Molecular Biology, led by the outstanding scientist Dr. Mirko Zimic. I chose this laboratory because they were developing interesting projects in Artificial Intelligence and Medicine. Thanks to my interest and my previous work, I was able to get the opportunity to work in their lab. There I started to help in projects where Machine Learning techniques were applied to make a faster and low cost diagnosis for several diseases such as anemia, autism and tuberculosis.

In parallel to my work at UPCH, I was doing projects independently. Thus, I had the opportunity to have one of my papers accepted in the LXAI workshop @ ICML 2019, and also to win a travel grant from the ICML organizers for ICML 2019. In that conference I was able to observe more closely the advances and applications of Machine Learning and Deep Learning, which motivated me to get more involved in the research and development of these interesting areas, besides that, I was able to establish contact with the LatinX in AI community and meet several latin guys involved in AI/ML.

Subsequently, some of my projects that I developed independently were submitted and accepted at conferences such as INTERCON and SIMBig. In addition, I submitted papers developed at UPCH to workshops at NeurIPS 2019 and ICML 2020. Most of those papers were successfully accepted and exhibited. I also had the opportunity to win a travel grant from LXAI for NeurIPS 2019, which helped me a lot to learn about the state of the art of ML/AI and also to expand my network.

A few months later, I won a Fondecyt-Peru grant to do an internship in Italy, in the laboratory of Dr. Lamberto Ballan, at the University of Padua. The work consisted in developing an automatic segmentation algorithm for tuberculosis cords in order to have a faster diagnosis. This work was successfully developed and the results were accepted for an oral presentation at the ML for Global Health workshop at ICML 2020. After finishing this internship I was accepted and won a scholarship to the Pi School of AI, where we had lectures on AI and developed a project for a company, the project was related to automatic generation of text summaries using BERT models. This experience helped me to see the differences between developing research projects and developing projects in a company, both of which present great challenges.

Since I had already published in several local conferences and workshops of major ML/AI conferences, I applied and was accepted to review papers for ICML/NeurIPS/ICLR/CVPR workshops, and at the same time I was invited to review papers for local conferences. This helped me to strengthen my ability to critically analyze research articles. Then I had the opportunity to participate in important summer schools such as the Lisbon Machine Learning School (LxMLS) 2020 and the Cornell, Maryland, Max Planck Pre-doctoral School (CMMRS) 2020, where I learned a lot about AI/ML fundamentals and the state of the art. At the end of 2020, I was accepted and obtained a scholarship for the 'Program in Data Science & Global Skills' which is organized by Aporta and MIT. I am currently in the program, which lasts a year and a half. There we are learning more about the fundamentals along with hands-on labs on Statistics, Probability, Machine Learning and Deep Learning. As part of this program we will also develop a Data Science project with an NGO.

Now, in 2021, I am still involved in ML/AI projects at Cayetano Heredia University. I am also participating in a project for an OpenCV competition where we passed the first stage, and at the same time I am waiting to participate in summers schools that accepted me such as the Eastern European Machine Learning Summer School (EEML) 2021, the Nordic Probabilistic AI School (ProbAI) 2021, among others.

I hope what I have shared here can help you as a small reference on how to get involved in AI/ML research and development, you can visit my personal website www.dennishnf.com for more information.


References

[1] Universidad Nacional de Ingeniería (UNI).

[2] Universidad Peruana Cayetano Heredia (UPCH).

[3] UNESCO/Poland Co-Sponsored Fellowship Programme in Engineering.

[4] LatinX in AI (LXAI).

[5] Pi School of Artificial Intelligence.

[6] Advanced Program in Data Science & Global Skills.

[7] International Conference on Machine Learning (ICML).

[8] Conference on Neural Information Processing Systems (NeurIPS).

[9] International Conference on Learning Representations (ICLR).