Fusemachines Nepal succesfully hosted a talk program on Deep Learning on 17th October 2017 at the Fusemachines Nepal Hall, Hattisar, Kathmandu. Entrance to the event was free, and the targeted participants were software engineers, engineering students, and AI enthusiasts.

The main speaker of the program DR. Steven J. Rennie introduced Deep Learning (DL), its brief history, its impact in the field of Machine Learning, and its challenges, during the one-hour long talk program.

The event was opened by Sameer Maskey, CEO and founder of Fusemachines. He gave a brief introduction about himself, the company and Steven before handing the mic over to Steven. Before starting the main presentation, Steven talked about his 20 years of experience in the field of Deep Learning and Machine Learning.

Steven started the presentation by stating that “everything has been totally revolutionized by DL. DL has remarkably improved the field of Machine Learning and Computer vision through accuracy improvements in computer vision problems such as object recognition and detection.”

He further went on to discuss how DL is actually improving computer vision. He gave a brief glimpse of the history and evolution of DL. Steven then laid out some core concepts of DL like CNN, RNN, Word Embedding and MS COCO. He also talked about the new practices of DL like Frame Randomization, Better Linearities, Better Initialization, Better Optimization and More General Models. According to him, the hottest area in DL right now is Adversarial Learning.

Towards the end, Steven shared his challenges he faced during his career. He talked about his work on the MS COCO and challenges in image captioning. He said that the key to solving the challenge was the use of Reinforcement Learning. After the presentation, Steven took some questions from the participants.

About the Speaker:

Dr. Steven Rennie was IBM Watson’s Team Lead (Multimodal Group) until recently. He joined Fusemachines in October as Director of Research in NY office. He is an expert in Deep Learning, Reinforcement Learning, Inference Algorithms, Graphical Models, Acoustic Modeling. He received his PhD from the University of Toronto in Electrical and Computer Engineering. He had been leading and coordinating R&D efforts on various activities related to IBM Watson Group. At Fusemachines he will be leading the global team of engineers and research scientists to invent new Artificial Intelligence algorithms. He is a leading authority on speech recognition, multi-talker ASR, image captioning and deep learning. He is also a member of New York Academy of Sciences, IEEE, SLTC, and ISCA.