AI For Enterprise 2021

At Philips, we believe the value of AI is only as strong as the human experience it supports. That’s why we combine the power of AI with deep clinical knowledge, gained through decades of experience in healthcare. We envision AI that is fully integrated into the workflows of healthcare professionals and people’s daily health routines – supporting them at every stage of the health continuum. With AI, we can help people take better care of their health and well-being, and enable healthcare professionals to do what they do best: prevent, diagnose, treat, and monitor. In this presentation we will highlight few important Ai principles that governs our AI Vision:

• Well-being – We design our solutions to benefit the health and well-being of individuals and to contribute to the sustainable development of society.
• Fairness – We develop and validate solutions using data that is representative of the target group for the intended use, and we aim to avoid bias or discrimination.
• Oversight- We design AI-enabled solutions to augment and empower people, with appropriate human supervision.
• Transparency- We disclose functions and features of our offerings that are AI-enabled, the validation process, and the responsibility for ultimate decision-taking.
• Robustness – We develop AI-enabled solutions that are intended to do no harm, with appropriate protection against deliberate or inadvertent misuse.

Vincent will walk through actionable takeaways for accelerating product development with AI that can be used at any organization. In his talk, Vincent will cover:

• A practical use case on the Natural Language Processing system that powers Shopify Inbox—a chat tool to manage customer conversations, create automated messages and get insights to focus on chats that convert
• How to build a data foundation to establish trust
• How to identify opportunities where only AI can solve at scale

• This presentation will cover what organisations need to understand and have in place in order to deploy ML models at scale while minimising risk.
• Joel Hodgson will discuss the common challenges faced by organisations when scaling ML model deployment and how to overcome them. He will discuss different use cases at different stages of ML maturity, as organisations move into deploying explainers and monitoring alongside their models.
• Sean Greaves will then walk through the creation, deployment, monitoring and management of a ML model using a real life use case and covering a range of practical tasks including training models on pre-processed datasets, deploying model artefacts with Seldon Deploy and using explainers and outlier detection to gain insight into model decisioning.

• What is AI bias, and is it fair? 
• Discussing liability and security 
• Who benefits from AI, and could it result in a decrease in human-human interaction and employment issues? 

• Developing Machine-Learning models is a very complex process. Regardless of industry and domain, developing a production-ready model that brings business value may take months. At H2O.ai, we have made it our mission to help companies around the world and across industries on their journey with AI adoption. We believe that automating Machine-Learning will let you develop better models faster.
• Speed and accuracy is not everything though. Trusting your model is equally important. With Automated Machine Learning Explainability, you can gain insight into the factors that drive your model, analyse its fairness and robustness, and get explanations for every prediction that you make.
• During this session, we will take a look at how Automated Machine Learning reduces the time to maximise business value. We will also discuss why it is important to have Machine Learning Explainability as an integral part of the overall process.

• What are the ethical dilemma’s with the development and deployment of AI
• To what extent do frameworks, regulations and tools provide meaningful guidance?
• How can or should organizations deal with these dilemma’s/The organizational dimension of responsible data science

Graphs are everywhere – everything is more connected every day and we increasingly struggle to make sense of the indirect connections that surround us, our businesses and our societies. Neo4j has been chipping away at this problem for decades and has created a Graph Platform that allows people to make sense of those connections. In this talk, we will explain

• How that platform works
• Why graph data science matters to reveal unknown insights from the data
• Demonstrate how the ICIJ used it to uncover yet another set of shady business practices, in the Pandora Papers

• Develop reliable AI products and processes
• Don’t waste your time, get things right from the start
• How to get good data; use cases from healthcare, automotive and other industries
• Global Scale through consistent data quality

• How to move your project from “experimentation” to “live”  
• R&D budget signoff – understanding why you need AI 
• Turning AI into ROI – what does value mean to your business? 
• Top tips for moving your pilot to the next stage 

• Why is it important to consider sustainability when looking at digital services and infrastructure
• What are the key considerations and opportunities
• How Bulk helped a UK based Fintech deliver a sustainable AI platform and significantly reduce cost at the same time

• Bol.com was the first online bookstore of Europe in 1999
• 20 years later, bol.com is the largest online retail platform in the Benelux, selling millions of products in dozens of product categories
• As one can expect, with the growth of bol.com, the number of fraud cases is increasing. In addition, fraudulent behaviour is constantly changing
• To stay ahead in this cat-and-mouse game the Fraud and Risk department of bol.com aims to proactively prevent fraud from happening on the platform
• How did the Fraud and Risk department move from a rule-based only reactive approach to a more proactive data-driven approach?

• Fraud, especially its dynamic nature, is a major area of concern requiring significant time and resources to isolate from an enormous volume of transaction data.  
• We have developed an innovative new composite AI based solution that combines graph-rule-based with graph-supervised-learning coupled with explainability to address this problem.  
• The talk is based on a real world Graph AI (GAI) project undertaken by Larus, Inc. and Fujitsu Research. The project evaluated the incorporation of GAI technology from Fujitsu (Deep Tensor) into an existing rules-based credit card fraud detection application developed by LARUS. 

• What is demand forecasting and how does it work?
• How did a leading DIY retailer utilize it to make more profit?
• How can you utilize it for your own company?