Session: Optimizing Machine Learning-Based Detection Systems
Detection systems are one of the industry's most profitable and widely used types of machine learning solutions. Therefore, its proper deployment is of extreme importance for a wide range of applications. Detection systems face several challenges when the data presents a highly imbalanced class distribution. In this talk, Gissel Velarde will share seven key points to achieve optimal performance in real-world detection problems for binary classification. She will describe relevant tasks that include understanding the business use case, exploring the data, exploring previous solutions, preparing the data, optimizing hyperparameters, evaluating performance, and finally, getting feedback from the end users.
Bio
Dr. Velarde is a Senior Expert Data Scientist at Vodafone. She earned a Ph.D. in Computer Science and Engineering from Aalborg University in Denmark in 2017, an institution recognized as the best university in Europe and fourth in the world in engineering according to the US News World Ranking and the MIT 2018 ranking. She holds a master's degree in Electronic Systems and Engineering management from the University of Applied Sciences of South Westphalia, Soest in Germany, thanks to a DAAD scholarship, and holds a Licenciatura's degree in Systems Engineering from Universidad Católica Boliviana, recognized as the third best university in Bolivia according to the Webometrics Ranking 2020. She has over 20 years of experience in engineering and computer science. She worked in industry and academia, and received several awards and recognition as an engineer, a computer scientist, and a pianist. She was a research member in the European Commission's project "Learning to Create", and a PhD Fellow at Aalborg University. She worked as an Engineer for Miebach Gmbh, Hansa Ltda, SONY Computer Science Laboratories, Moodagent, among others. Velarde developed machine learning and deep learning algorithms for classification, structural analysis, pattern discovery, and recommendation systems. In 2020, Velarde was named: Notable Woman in the field of research by the International Society for Music Information Retrieval (ISMIR). She was also selected internationally as one of 120 technologists by the Top Women Tech 2019 & 2020 summit in Brussels. In 2021, Velarde received a teaching award from Universidad Privada Boliviana for her lecture on Selected Topics in Artificial Intelligence. She received twice a Vodafone Star Award for her contributions.