It's time to head back to Campus London for another Applied AI meetup, everyone! We're excited to welcome to the stage Stathis Vafeias from AimBrain, Andrew Clegg a machine learning consultant, and NStack.
NEW! We're using Eventbrite for registrations & tickets so please head over to our Eventbrite page (https://www.eventbrite.co.uk/e/applied-ai-meetup-tickets-35721203080) to secure your place.
6:30pm - Beer, food and chat
7:00pm - Stathis Vafeias
7:30pm - Andrew Clegg
8:00pm - Nick Pollard and Mark Szepieniec
AutoML: time to evolve
Building deep learning models has been portrayed as playing with lego blocks. The success of the model highly depends on the configuration of its blocks. Hand designing new architectures for every variation of a problem can be time consuming. In this talk I will talk about using evolution to explore the structure space of a neural network.
Stathis holds a PhD in Robotics from Edinburgh University. Currently he leads the machine learning at AimBrain, where he works on deep learning models for mobile biometric authentication. Before joining AimBrain, he was a research engineer at Toshiba Medical Visualization Systems.
Learning representations for unordered item sets
RNNs and their more sophisticated cousins (GRUs, LSTMs) have proven to be "unreasonably effective" at learning from sequence data, although they can be tricky and expensive to train. But what if your data consists of unordered sets or bags of objects, or the data is ordered but the predictive value of that ordering is marginal? Deep Averaging Networks (Iyyer et al 2015) provide a cheap and effective way to learn task-oriented embeddings from this kind of data. This talk will introduce DANs and demonstrate their use on a retail dataset.
Andrew is a machine learning engineer with an academic background in NLP and IR, and a career that's weaved its way through biomedical science, social media, online music, educational technology and e-commerce. These days he's into recommendations, personalization, search, cats, and food.
Functional Data Science and Algebraic Infrastructure
Nick Pollard and Mark Szepieniec
Functional programming concepts such as composition, immutability, and type-safety allow software developers to rapidly build reliable, reusable, and correct systems. These approaches can also be used to solve many of the problems that data scientists face in the wild -- such as sharing and reusing models, connecting them to data sources and third-party systems, and ensuring predictable behaviour in production.
In this presentation, we will explore how concepts such as composition, immutability, and type-safety can be used by data scientists to productionise and integrate models.
Nick Pollard is a Senior Software Engineer at NStack, and previously led development of a high-performance, typed streaming system in Scala/Scalaz at Scotiabank. Mark Szepieniec, Data Scientist at NStack, was previously a Data Scientist at ONZO, and is completing his PhD in Computational Physics.
For tickets, please go here: Applied AI meetup Eventbrite (https://www.eventbrite.co.uk/e/applied-ai-meetup-tickets-35721203080).