Avatar

Emmanuel Agbeli

Data Scientist/ML Researcher

Biography

I am a Machine learning Researcher and Engineer. I am most interested in works that combines multple discipline and of machine learning and its application. I shape my efforts around using ML reasoning to address key challenges in real-life scenario. He holds double masters from the African Masters of Mathematical Sciences, one in Machine Intelligence sponsored by Google and Meta and the other in Big Data and Financial Mathematics. His research work in both programs was on Self-supervised domain adaptation and ML in Game theory. He had his bachelor studies in Statistics from the Kwame University of Science and Technology, KNUST.

Research Interests

  • Deep learning
  • Representation learning: Vision and language
  • Probabilistic modeling and statistical inference
  • Reinforcement learning

Education

  • MSc. Machine Intelligence

    Africa Masters of Machine Intelligence(AMMI-Rwanda)

    Supervisor: Dr. Naila Murray

  • MSc. in Mathematical Sciences, 2019

    Africa Masters of Mathematical Sciences(AIMS-Senegal)

    Supervisor: Dr. Arne Ring

  • BSc. in Statistics, 2016

    Kwame Nkrumah University of Science and Technology(KNUST)

    Supervisor: Dr. Nana Kena Frimpong

News

  • [Dec 2021] I would be moderating for this year's 2021 Neurlips Black-in-AI workshop (Volunteering)

  • [Jul 2021] I would be volunteering as a tutor for this year's Neuromatch Academy Deep Learning Summer School(NMA DL).

  • [March 2021] Officially starting a Data Scientist role at Yemaachi Biotechnology. The first Africa Cancer Research lab based in Accra-Ghana.

  • [Dec 2020] I am honored to give a presentation on representation learning at AI-Ghana meet-up. The talks is on brief Introduction to clusterization of unlabeled and quantization of datasets such as images. Repos

  • [Dec 2020] I was accepted into the Neural Information Processing System Conference 2020. This is annual gathering of researchers in the area of artificial intelligence and it application humanity

  • I recently participated in the Kaggle competition on NLP for tweet disaster in detecting whether a tweet is about is fake or real.

Work experience

Yemaachi Biotechnology company

Data Scientist
March 2021 – March 2022.

Dataware Tech Ghana

Part-time position: ML Engineer
Oct 2020 – Feb 2021.

International Crop Research Institute for the Semi-Arid Tropics (ICRISAT)

Research Apprentice/Data Scientist
June 2019 - Sep 2019.

Saham Insurance ,Kigali-Rwanda

Data Scientist-Intern
Jun 2018 – Dec 2018.

Local government, Tema Metropolitan- Ghana

IT Assistant,Intern
Jun 2014-Aug 2014.

Teaching experience

Neuromatch Academy Summer School

Teaching assistant(Volunteering)
Jul. 2021 – Aug 2021.

University of Energy and Natural Resources, Sunyani-Ghana

Teaching and Research Assistant
Sep 2016 – Aug 2017.

Projects

All

CGIAR Computer Vision for Crop Disease

Wheat rust is a devastating plant disease that affects many African crops, reducing yields and affecting the livelihoods of farmers and decreasing food security across the continent. The disease is difficult to monitor at a large scale, making it difficult to control and eradicate. I had to implement deep learning architecture for detecting the type of disease.

Code

Natural Language Processing with Disaster Tweets

Twitter has become an important communication channel in times of emergency. The ubiquitousness of smartphones enables people to announce an emergency they’re observing in real-time. Because of this, more agencies are interested in programatically monitoring Twitter (i.e. disaster relief organizations and news agencies). In order to address this problem, I intend to leverage on ML to assist in detecting real or fake tweets.

Code

ASR system for low-sourced language(Twi)

In-class project with the aim to build Automatic Speech Recognition for low-sourced language such as Twi. This is to translate the speech from English using Temporal Classification Connectionist. Source code

Code

Enzyme Prediction

All enzymes are made of one or more chains of amino acids, which determine their structure, behaviour, and interactions with other enzymes and molecules. That means it should be possible to predict the protein’s function and behaviour given just the amino acid sequence. I implemented a machine learning algorithm to classify the labels of amino acid sequence.

Code

Skills

Programming

  • Python

  • Matlab

  • R-software

Tools

  • Git

  • MySQL

  • Latex/Microsoft suits

Libraries

  • Pytorch

  • Tensorflow

  • NLTK/SpaCy

  • OpenCV

Contact

Feel free to contact me