Based on some feedback from our first Meetup for the IBM PowerAI London group, we will cover some use cases for AI and Deep Learning in this session.
In this talk we will discuss the role of human-in-the-loop analytics that incorporate machine learning or reasoning capabilities. Our emphasis is on how machine learning can enhance and engage the experience of analysts working in domains such as cyber security and support complex decision-making / -planning. We shall outline how the increasingly significant challenges of data volume, variety and veracity impede progress within these fields and how practical, well-designed AI techniques can make a positive difference and these problems more tractable. Relativistic anomaly scoring systems based on tensors, bespoke unsupervised clustering techniques and manifolds in high-dimension datasets; Use of container-based machine reasoning in scenario planning, using controlled natural language that is human intuitive and machine readable; and Application of machine learning in generative signature models applied to advanced cyber security and self-healing networks. In parallel with these examples we shall describe a model of computing in which runtimes are called and applied based on assessing their 'best-fit' for a given problem. This speaks to a future in which classical machines are used in concert with hybrid or highly optimised specialist platforms, universal quantum computers, non von Neumann architectures, approximate and stochastic computing paradigms.
Leigh is a Computer Scientist and Technical Leader within the Emerging Technology group of IBM Research. Based at the Hursley laboratories (UK) his specialisms are information security, artificial intelligence / machine learning and scientific computing. Much of Leigh's current work relates to experimental computer science; specifically the design and creation cognitive agents that bolster existing network defence technologies, working alongside human analysts to improve visibility and understanding of complex systems.
H2O Driverless AI is an artificial intelligence (AI) platform that automates some of the most difficult data science and machine learning workflows such as feature engineering, model validation, model tuning, model selection and model deployment. It aims to achieve highest predictive accuracy, comparable to expert data scientists, but in much shorter time thanks to end-to-end automation. Driverless AI also offers automatic visualizations and machine learning interpretability (MLI). Especially in regulated industries, model transparency and explanation are just as important as predictive performance.
Jo-fai (or Joe) is a data scientist at H2O.ai. Before joining H2O, he was in the business intelligence team at Virgin Media in UK where he developed data products to enable quick and smart business decisions. He also worked remotely for Domino Data Lab in the US as a data science evangelist promoting products via blogging and giving talks at meetups. Joe has a background in water engineering.
Data Science Experience is an IBM tool set that allows Data Scientists to create and share their Jupyter notebooks in a more collaborative and manageable way. With support for common data science environments including Python, Scala and R, as well as the majority of Machine Learning and Deep Learning frameworks. In this environment Data Scientists can work with a single source of data, exploring and manipulating it using the best of breed tools and frameworks in common usage.
Mark will showcase some of the benefits of the DSX Environment for Data Scientists, by walking the audience through a Notebook that looks at a data set of (anonymised) credit card transactions. This looks at traditional data management strategies, and also how to then make use of Machine Learning and Deep Learning frameworks to analyse this data, find patterns within that data - and ultimately build an algorithm to spot fraudulent transactions in the future.
Mark started his journey as a Support Programmer with Informix 4GL for a small, independent software vendor. He then moved to a Consultancy role in Network Performance Management in the Telecommunications Sector working for another independent software vendor. From there, Mark worked his way up to a Technical Account Manager before being part of a software acquisition into IBM back in 2007.