There is an evolution in software development. Traditionally built by programmers who evaluate any and all possible situations, and then write rules to manage them, we now have applications created by machine learning algorithms. We are migrating from programmatically proving things correct and now are running experiments. Software development is becoming more of a natural science. This shift in the way we build software is only the beginning of new and exciting possibilities — and challenges.
CrowdFlower has seen customers from a broad array of industries, from financial institutions to retailers, push video and image tasks through the CrowdFlower platform. Watch as CrowdFlower co-founders, Lukas Biewald and Chris Van Pelt, walk through the latest set of CrowdFlower Computer Vision tools, as well as show use cases of how our customers are using them.
Kaggle has seen more machine learning models than any other company. Anthony Goldbloom, co-founder and CEO of Kaggle, shares lessons his company has learned from the more than 2 million machine learning models that have been submitted to Kaggle competitions.
Digital creativity is important to Adobe. Last year, Adobe introduced Adobe Sensei, a unified machine learning and AI framework that can harness trillions of data content and assets. Scott Prevost reviews the work and results from using Adobe Sensei.
At a talk five years earlier, Robert presented a topic entitled “Where’s My Talking Robot”, a conversation which focused on the work done to bring conversational AI to what is now today a reality. In this talk, he speaks to how are we interfacing with AI, and what our expectations of interfacing with AI will be in the coming future.
In Barney Pell’s talk, he speaks to the multiple issues related to the quality, robustness and usability which often affect the building of a successful AI application. In addition, finding the perfect balance of humans and automation, trust, evolution and economics can introduce even more challenges. All can be solved if you first ask the right questions.
How do we best use human judgement and do items in different datasets represent the same thing? In Toby’s talk, he focuses on how “human-in-the-loop” (the approach of using real people to collect, clean and label your dataset for better training data for your machine learning models) can impact and improve the algorithmic results in real time.
The way people search for things has changed and has gone beyond just typing in words and pulling up results. It is more about creating an ultimate digital assistant which can leverage natural language processes that can extract what users are saying and interpret what they need. In this talk, the team from Ozlo speaks to how they seek to create an integrated knowledge platform for intelligent interfaces.
LexisNexis, creator of Legal Answers, a legal questioning and answering system built for lawyers using machine learning and other artificial intelligence technologies, enables lawyers to quickly find straight forward answers to simple legal questions that would normally take a long time.
Planet Labs has a goal to image every square meter of earth, everyday, to drive meaningful change. They do this by radically designing their satellites so they can launch hundreds of them into space for the purpose of mapping. In his talk, Sid Dixit talks to how this massive amount of data is helping change the world.
Ryan Keiser, Head of Machine Learning at Descartes Labs, a startup based in New Mexico, presents how Descartes Labs works to drastically improve access to satellite data and to build a living atlas of the world.
Spotify considers music emotional. It affects your mood and the overall impact on humans is what draws interest by Spotify. Hear how datasets, and when applied to machine learning, can collectively improve and enhance the music we choose to listen to.
Paul Ogilvie, Engineering Manager on the machine learning algorithm team at LinkedIn, discusses how LinkedIn personalized job recommendations on the LinkedIn platform use machine learning.
According to Xuedong Huang, Technical Fellow at Microsoft, if you want to do anything useful with AI, you need three components – a big computer, a powerful deep learning algorithm and massive amounts of data. AI is soon to be everything and Microsoft wants to help organizations become more productive with AI. In this talk we learn about Microsoft’s Cognitive Toolkit (CNTK) and the impact it is trying to make on the way we use AI.
James Cham and Shivon Zills, partners at Bloomberg Beta, a venture fund which invest in early-stage technology companies, give an overview of the current machine learning landscape.
Monica Rogati, an early adopter of the CrowdFlower platform, discusses the past, present and future of data science by reflecting on the evolution of this fast growing field.
HERE Technologies, the leading mapping technology company in the world, talks to creation of their HERE Reality Index, a geospatial dataset that seeks to gather information about all of the physical digitally connected objects in the world and their locations.
Presented by the co-founder and CTO of Clara Labs, Michael Akilian, his talk discusses the reason why he founded Clara Labs and how using machine learning and CrowdFlower was able to solve one of business’ most time consuming efforts.
AI has proven to be a valuable resource in the analysis of images relevant to the health and medicine industry. In this talk, Dan Ruderman presents the increase in the use of digital clinical imagery and how AI solutions are helping to improve the accuracy and efficiency of medical research and diagnoses.
Richard Socher, Chief Scientist at Salesforce, presents solutions to how to predict previously unseen words at test time, single input and output encoding for words, growing a single model for many tasks and how to use a single end-to-end trainable architecture for question answering.