Deep neural network facial recognition models like OpenFace, VGGFace are achieving amazing results on facial landmark detection, head pose estimation, facial action unit recognition, and eye-gaze estimation.
These models combined with the right training data, can be used by banks, telcos and other credit providers to verify the identity of their consumers. For example to compare a selfie image against the picture of their government issued ID. Adding OCR can allow further verification of personal information, such as their name, date of birth, ID number, etc.
We will compare various model architectures, embeddings, data augmentations, training datasets used for facial recognition applications.
Arshak Navruzyan is a machine learning focused product manager. He founded Fellowship.AI applied machine learning fellowship program and is the Chief Technology Officer at Sentient Technologies. Arshak has delivered AI solutions for multi-billion dollar quantitative hedge funds, numerous venture funded startups and some of the largest telecoms in the world. Arshak has been in technology leadership roles at Argyle Data, Alpine Data Labs, Endeca/Oracle.
Sharath Kalkur is a Masters graduate from University of Illinois at Chicago, majoring in Electrical and Computer Engineering. He specialized in the field of Machine Learning and Data Sciences, Computer Vision and Processor Technology. He has previously worked as a Machine Learning Intern and AI Architect at TrueMedicines Inc. and a Machine Vision Engineer at Waec, LLC. Sharath is now a Machine Learning Fellow at Fellowship.AI. He has worked closely on Launchpad.AI's Identity over the course of his fellowship.
Machine Learning has made our lives more productive from hailing a ride via Uber’s advanced ML-driven rider and driver matching, or Google Now predicting information you’d need before you need it. Machine learning has also made our lives safer allowing people to rent strangers’ houses via Airbnb or reducing the risk of fraud during online purchases. Recent advances in deep learning have brought more new technologies within our reach including self-driving cars, machine translation, predicting weather several years ahead, automated stock trading and more! In this track, come hear from practitioners about some interesting applications of machine learning and recent practical advances in deep learning.
The track host is looking for additional speakers, in particular deep learning practitioners or those with novel and practical applications of machine learning. If interested, please contact the track host, Soups Ranjan at [masked]
We invite you to attend the first annual Data Intelligence conference. This event aims to bring machine learning engineers and researchers together to share ideas and projects amongst both academic and professional practitioners in the field.
Conference start/end times:
Dates: Friday June 23 to Sunday[masked]
Location: McLean, Virginia (DC Metro Area)
Adversarial Machine Learning
Machine learning techniques were originally designed for environments in which the training and test data are assumed to be generated from the same (although possibly unknown) distribution and/or process. In the presence of intelligent and adaptive adversaries, however, this working hypothesis is likely to be violated.
Applying machine learning to use cases like fraud, security, anti-money laundering and know your customer (KYC) presents a unique set of challenges:
- Little or no labeled data
- Non-stationary data distributions
- Model decay
- Counterfactual conditions
This event is entirely devoted to understanding how modern machine learning methods can be applied to these adversarial environments. We will have hands-on workshops as well as talks by leading practitioners from industry and academia.
Register at http://conf.startup.ml/adversarial
Our friends @Quantopian are hosting their second annual conference, QuantCon 2016 in NYC on April 9, 2016. Stellar lineup including Dr. Emanuel Derman and Dr. Marcos López de Prado. 10% discount code for our community on tickets: StartupMLDiscount102016. Reserve your spot today:
Incremental learning tools such as Vowpal Wabbit (VW) and Sofia-ML can learn massive datasets by streaming the data through a fixed-size memory window. This means that they can learn a useful model from datasets that are much larger than the amount of system memory. Progressive validation (http://hunch.net/~jl/projects/prediction_bounds/progressive_validation/coltfinal.pdf) is at the heart of this learning approach. It forces the model to make a prediction before seeing the true label of the example, yielding a surprisingly reliable estimate of generalization error.
We'll do a quick walk through of VW and Sofia with some live demonstrations of model training and testing.
Speaker: Arshak Navruzyan, VP Product @ Argyle Data
My objective is to make distributed systems and machine learning accessible to any organization or individual that wants to transform the world through data. I am currently VP of Product Management at Argyle Data focused on petabyte-scale risk management applications using machine learning and Hadoop. Previously I held senior engineering and product management roles at Alpine Data Labs, Endeca and Oracle. I am a contributor to the Apache Accumulo project and the organizer of San Francisco Machine Learning Meetup group.