Program chairs Ben Lorica and Roger Chen open the first day of keynotes.
Software engineering of systems that learn in uncertain domains
Peter Norvig (Google)
Building reliable, robust software is hard. It is even harder when we move from deterministic domains (such as balancing a checkbook) to uncertain domains (such as recognizing speech or objects in an image). The field of machine learning allows us to use data to build systems in these uncertain domains. Peter Norvig looks at techniques for achieving reliability (and some of the other -ilities).
Why we'll never run out of jobs
Tim O'Reilly (O'Reilly Media)
There are many who fear that in the future, AI will do more and more of the jobs done by humans, leaving us without meaningful work. To believe this is a colossal failure of the imagination. Tim O'Reilly explains why we can't just use technology to replace people; we must use it to augment them so that they can do things that were previously impossible.
Artificial intelligence: Making a human connection
Genevieve Bell (Intel Corporation)
Genevieve Bell explores the meaning of “intelligence” within the context of machines and its cultural impact on humans and their relationships. Genevieve interrogates AI not just as a technical agenda but as a cultural category in order to understand the ways in which the story of AI is connected to the history of human culture.
Humanizing AI development: Lessons from China and Xiaoice at Microsoft
Lili Cheng (Microsoft)
Keynote by Lili Cheng
How AI is propelling driverless cars, the future of surface transport
Shahin Farshchi (Lux Capital)
Keynote by Shahin Farshchi
Ben Lorica (O'Reilly Media), Roger Chen (.)
Program chairs Ben Lorica and Roger Chen close the first day of keynotes.
Deep neural network model compression and an efficient inference engine
Song Han (Stanford University)
Neural networks are both computationally and memory intensive, making them difficult to deploy on embedded systems with limited hardware resources. Song Han explains how deep compression addresses this limitation by reducing the storage requirement of neural networks without affecting their accuracy and proposes an energy-efficient inference engine (EIE) that works with this model.
The future of AI
Oren Etzioni (Allen Institute for Artificial Intelligence)
Oren Etzioni offers his perspective on the future of AI, based on cutting-edge research at the Allen Institute for AI on projects such as Aristo and Semantic Scholar. This future reflects the institute's mission: AI for the common good.
How advances in deep learning and computer vision can empower the blind community
Anirudh Koul and Saqib Shaikh explore cutting-edge advances at the intersection of computer vision, language, and deep learning that can help describe the physical world to the blind community. Anirudh and Saqib then explain how developers can utilize this state-of-the-art image description, as well as visual question answering and other computer-vision technologies, in their own applications.
Deep reinforcement learning for robotics
Pieter Abbeel (OpenAI / UC Berkeley)
Pieter Abbeel explores deep reinforcement learning for robotics.
Verticals and applications
End-to-end learning for autonomous driving
Urs Muller (NVIDIA)
Urs Muller presents the architecture and training methods used to build an autonomous road-following system. A key aspect of the approach is eliminating the need for hand-programmed rules and procedures such as finding lane markings, guardrails, or other cars, thereby avoiding the creation of a large number of “if, then, else” statements.
High-level APIs for scalable machine learning
Martin Wicke (Google)
TensorFlow is a system for scalable machine learning. However, using raw TensorFlow and profiling, optimizing, and debugging large-scale models can be daunting for novice and expert users alike. Martin Wicke explores new APIs based on TensorFlow that aim to make building complex models easier and allow users to scale quickly.
Impact on business and society
Only humans need apply: Adding value to the work of very smart machines
Tom Davenport (Babson College, MIT)
The automation of decisions and actions now threatens even knowledge-worker jobs. Tom Davenport describes both the threat of automation and the promise of augmentation—combining smart machines with smart people—and explores five roles that individuals can adopt to add value to AI, as well as what these roles mean for businesses.
Practical AI product development
Hilary Mason (Fast Forward Labs)
Hilary Mason explores a framework for applied AI research, with a focus on algorithmic capabilities that are useful for building real-world products today. Drawing on real-world examples, Hilary outlines a system for thinking about which AI capabilities are ready to transition from pure research to applied products and how to make the transition from research paper to a working product.
Transforming your industry with cognitive computing
Guruduth Banavar (Cognitive Computing, IBM)
In the last decade, the availability of massive amounts of new data, the development of new AI techniques, and the availability of scalable computing infrastructure have given rise to a new class of machine capabilities we call cognitive computing. Guruduth Banavar offers an overview of the technological breakthroughs that are enabling this trend.
Benefits of scaling machine learning
Reza Zadeh (Matroid | Stanford)
Machine learning is evolving to utilize new hardware, such as GPUs and large commodity clusters. Reza Zadeh presents two projects that have benefitted greatly through scaling: obtaining leading results on the Princeton ModelNet object recognition task and matrix computations and optimization in Apache Spark.
Unlock the power of AI: A fundamentally different approach to building intelligent systems
Mark Hammond (Bonsai)
Mark Hammond explains how Bonsai’s platform enables every developer to add intelligence to their software or hardware, regardless of AI expertise. Bonsai’s suite of tools—a new programming language, AI engine, and cloud service—abstracts away the lowest-level details of programming AI, allowing developers to focus on concepts they want a system to learn and how those concepts can be taught.
The need for speed: Benchmarking deep learning workloads
Greg Diamos (Baidu), Sharan Narang (Baidu)
Greg Diamos and Sharan Narang discuss the impact of AI on applications within Baidu, including autonomous driving and speech recognition, offering a brief introduction to the challenges in training deep learning algorithms as well as the different workloads that are used in various deep learning applications.
How to make robots empathetic to human feelings in real time
Pascale Fung (The Hong Kong University of Science and Technology)
Pascale Fung describes an approach to enable an interactive dialogue system to recognize user emotion and sentiment in real time and explores CNN models that recognize emotion from raw speech input without feature engineering and sentiments. These modules allow otherwise conventional dialogue systems to have “empathy” and answer users while being aware of their emotion and intent.
Building and applying emotion recognition
Anna Roth (Microsoft), Cristian Canton (Microsoft Technology and Research)
Anna Roth and Cristian Canton walk you through building a system to recognize emotions by inferring them from facial expressions. Cristian and Anna explain how they trained their emotion recognition CNN from noisy data and how to approach labeling subjective data like emotion with crowdsourcing before showing a demo of this work in action, as it is exposed in Microsoft’s Emotion API.
Interacting with AI
Deeply active learning: Approximating human learning with smaller datasets combined with human assistance
Christopher Nguyen (Arimo), Binh Han (Arimo)
Natural-language assistants are the emergent killer app for AI. Getting from here to there with deep learning, however, can require enormous datasets. Christopher Nguyen and Binh Han explain how to shorten the time to effectiveness and the amount of training data that's required to achieve a given level of performance using human-in-the-loop active learning.
Interacting with AI
Probabilistic programming for augmented intelligence
Vikash Mansinghka (MIT), Richard Tibbetts (MIT)
The next generation of AI systems will provide assisted intuition and judgment for everyday people trying to collaboratively solve hard problems. Vikash Mansinghka and Richard Tibbetts explore how AI will be used on problems like malnutrition, public health, education, and governance—complex, ambiguous areas of human knowledge where data is sparse and there are no rules.
Impact on business and society
The future of natural language generation, 2016–2026
Robbie Allen (Automated Insights, Inc.)
Natural language generation, the branch of AI that turns raw data into human-sounding narratives, is coming into its own in 2016. Robbie Allen explores the real-world advances in NLG over the past decade and then looks ahead to the next. Computers are already writing finance, sports, ecommerce, and business intelligence stories. Find out what—and how—they’ll be writing by 2026.
Impact on business and society
The new artificial intelligence frontier
Babak Hodjat (Sentient Technologies)
Babak Hodjat discusses the progress in AI, diving into how AI can offer unique solutions in verticals such as investment, medical diagnosis, and ecommerce. Babak details how using massively scaled distributed evolutionary computation, mimicking biological evolution, allows an AI to learn, adapt, and react faster to provide customers with the answers and decisions they need.
AI on IA
Vin Sharma (Intel)
Vin Sharma explores how Intel is investing in artificial intelligence and using open source software platforms, frameworks, and libraries, as well as Intel's hardware to advance the future.
Deploying AI-based services in the data center for real-time responsive experiences
Sanford Russell (NVIDIA)
In the new era of artificial intelligence, every organization must examine how to extract intelligence from its data using deep learning. Sanford Russell explores how NVIDIA GPUs are deployed today to accelerate deep learning inference workloads in the data center.
Growing up: Continuous integration for machine-learning models
Zachary Hanif (Capital One)
Developing and validating frequently updated models is core to professional data science teams. Zachary Hanif discusses the adaptation of CI tools and practices to solve model governance and accuracy tracking concerns in a complex environment with adversarial and temporal data complications.
Impact on business and society
What I learned by replacing middle-class manufacturing jobs with ML and AI
Eduardo Arino de la Rubia (Domino Data Lab)
Manufacturing in the United States is facing extreme pressures from globalization. Eduardo Arino de la Rubia synthesizes what he learned working side by side with the workers he was replacing with AI and ML, discussing their struggles, how they saw the technology the would take their jobs, the limitations of the technology, and what his real impact was in the face of globalization.
Deep learning: Modular in theory, inflexible in practice
Diogo Moitinho de Almeida (Enlitic)
The high-level view of deep learning is elegant: composing differentiable components together trained in an end-to-end fashion. The reality isn't that simple, and the commonly used tools greatly limit what we are capable of doing. Diogo Almeida explains what we can do about it and offers a practical attempt at a deep learning library of the future.
A peek at x.ai’s data science architecture
Angela Zhou (x.ai)
In any human-machine interaction, you need a dialogue model: the machine must understand and be able to respond appropriately. Angela Zhou discusses x.ai's AI personal assistant, Amy Ingram, who schedules meetings for you, focusing on the way x.ai has approached both understanding and responding.
Break: Morning coffee service Sponsored by Capital One
Break: Morning Break
Break: Lunch Sponsored by Intel
Break: Afternoon Break
Come enjoy delicious snacks and beverages with fellow O'Reilly AI attendees, speakers, and sponsors.