Deep learning models have become an increasingly preferred option for biometric recognition systems, such as speaker recognition. SincNet, a deep neural network architecture, gained popularity in speaker recognition tasks due to its parameterized sinc functions that allow it to work directly on the speech signal. The original SincNet architecture uses the softmax loss, which may not be the most suitable choice for recognition-based tasks. Such loss functions do not impose inter-class margins nor differentiate between easy and hard training samples. Curriculum learning, particularly those leveraging angular margin-based losses, has proven very successful in other biometric applications such as face recognition. The advantage of such a curriculum learning-based techniques is that it will impose inter-class margins as well as taking to account easy and hard samples. In this paper, we propose Curricular SincNet (CL-SincNet), an improved SincNet model where we use a curricular loss function to train the SincNet architecture. The proposed model is evaluated on multiple datasets using intra-dataset and inter-dataset evaluation protocols. In both settings, the model performs competitively with other previously published work. In the case of inter-dataset testing, it achieves the best overall results with a reduction of 4\% error rate compare to SincNet and other published work.

Reem Mahmoud is the Co-founder and Education Lead at Zaka. Reem's passion for Machine Intelligence and education is the main driver for her role at Zaka. She is currently pursuing a Ph.D. degree at the American University of Beirut (AUB) in Electrical & Computer Engineering where her research is in the area of Machine Intelligence with a focus on learning from limited data in the IoT and sensing applications. Her research interests also include digital signal processing, optimization methods, and computer vision. Reem graduated with a B.S. in Electrical Engineering with high distinction from Alfaisal University in Riyadh, Saudi Arabia in 2015 and received her M.E. in Electrical & Computer Engineering from AUB in 2017 where her thesis was about designing accurate personalized user models from sensing data.

Department of Electrical & Computer Engineering (ECE) welcomed Professor Nicolas Papernot as its newest faculty member this fall. He joins ECE from Pennsylvania State University after spending a year at Google Brain as a research scientist. We sat down with Professor Papernot to hear about his research, why he chose ECE at U of T and asked him what advice he had for the class of 2T3. You joined us after spending a year at Google. Can you tell us a bit about your academic history?

From left: Professor Brendan Frey (ECE), U of T Engineering alumnus and CEO of Jupiter Networks Rami Rahim, Dean Cristina Amon and ECE Chair Professor Farid Najm. Your computer can identify you in photos, sort your email and keep your money safe online, but like anyone starting a new job, first it needs to learn how. The field of machine learning has exploded in recent years, and researchers are now developing software systems that can "learn" by exposure to training data sets and make "decisions" about real-world problems in ways that approach or exceed human abilities. On Thursday, June 23, alumni of the University of Toronto's Faculty of Applied Science & Engineering, current students and friends of the Faculty gathered at a BizSkule Speaker Series event in California to learn about the next frontier of machine learning from one of the field's experts, Brendan Frey, a professor in The Edward S. Rogers Sr. Hosted by Rami Rahim (ElecE 9T4), CEO of Juniper Networks at the company's headquarters in Silicon Valley, BizSkule showcases engineering leadership by inviting keynote speakers and industry panellists to share their experiences and insights on key issues and hot topics. Professor Frey and his team are working on closing what he calls the genotype-phenotype gap: understanding how a person's genomic make up influences their physical characteristics and traits.