Mar 15–18, 2020

Schedule: AI sessions

Add to your personal schedule
9:00am12:30pm Monday, March 16, 2020
Location: LL21 C
Sourav Dey (Manifold), Alex Ng (Manifold)
ML engineers work at the intersection of data science and software engineering—that is, MLOps. Sourav Dey and Alex Ng highlight the six steps of the Lean AI process and explain how it helps ML engineers work as an integrated part of development and production teams. You'll go hands-on with real-world data so you can get up and running seamlessly. Read more.
Add to your personal schedule
11:00am11:40am Wednesday, March 18, 2020
Location: 210 E
Martin Förtsch (TNG), Thomas Endres (TNG Technology Consulting GmbH)
Imagine looking into a mirror, but not seeing your own face. Instead, you're looking in the eyes of Barack Obama or Angela Merkel. Your facial expressions are seamlessly transferred to the other person's face in real time. Martin Förtsch and Thomas Endres dig into a prototype from TNG that transfers faces from one person to another in real time based on deepfakes. Read more.
Add to your personal schedule
11:00am11:40am Wednesday, March 18, 2020
Location: 210 B
Colin Spikes (Algorithmia)
ML is advancing rapidly, but only a few contributors focus on the infrastructure and scaling challenges that come with it. Colin Spikes explores why ML is a natural fit for serverless computing, a general architecture for scalable ML, and common issues when implementing on-demand scaling over GPU clusters. He provides general solutions and describes a vision for the future of cloud-based ML. Read more.
Add to your personal schedule
11:00am11:40am Wednesday, March 18, 2020
Location: 210 F
Devices discover their way around the network and proxy the intent of the users behind them; leveraging this information for behavior analytics can raise privacy concerns. A selective use of embedding models on a crafted corpus from anonymized data can address these concerns. Ramsundar Janakiraman details a way to build representations with behavioral insights that also preserves user identity. Read more.
Add to your personal schedule
11:50am12:30pm Wednesday, March 18, 2020
Location: 210 E
Meghana Ravikumar anchors on building an image classifier trained on the Stanford Cars dataset to evaluate fine tuning, feature extraction, and the impact of hyperparameter optimization, then tune image transformation parameters to augment the model. The goal is to answer: how can resource-constrained teams make trade-offs between efficiency and effectiveness using pretrained models? Read more.
Add to your personal schedule
11:50am12:30pm Wednesday, March 18, 2020
Location: 210 B
Nick Pinckernell (Comcast)
With model serving becoming easier thanks to tools like Kubeflow, the focus is shifting to feature engineering. Nick Pinckernell reviews five ways to get your raw data into engineered features (and eventually to your model) with open source tools, flexible components, and various architectures. Read more.
Add to your personal schedule
11:50am12:30pm Wednesday, March 18, 2020
Location: 210 F
Krishna Gade (Fiddler Labs)
Krishna Gade outlines how "explainable AI" fills a critical gap in operationalizing AI and adopting an explainable approach into the end-to-end ML workflow from training to production. You'll discover the benefits of explainability such as the early identification of biased data and better confidence in model outputs. Read more.
Add to your personal schedule
1:45pm2:25pm Wednesday, March 18, 2020
Location: LL20D
Nicola Corradi (DataVisor)
Fraudulent attacks such as application fraud, fake reviews, and promotion abuse have to automate the generation of user content to scale; this creates latent patterns shared among the coordinated malicious accounts. Nicola Corradi digs into a deep learning model to detect such patterns for the identification of coordinated content abuse attacks on social, ecommerce, financial platforms, and more. Read more.
Add to your personal schedule
1:45pm2:25pm Wednesday, March 18, 2020
Location: 210 E
Stephan Erberich (University of Southern California), Kalvin Ogbuefi (Children's Hospital Los Angeles), Long Ho (Children's Hospital Los Angeles)
Annotating radiological images by category at scale is a critical step for analytical ML. Supervised learning is challenging because image metadata doesn't reliably identify image content and manual labeling images for AI algorithms isn't feasible. Stephan Erberich, Kalvin Ogbuefi, and Long Ho share an approach for automated categorization of radiological images based on content category. Read more.
Add to your personal schedule
1:45pm2:25pm Wednesday, March 18, 2020
Location: 210 B
Fidan Boylu Uz (Microsoft), Mario Bourgoin (Microsoft), Gheorghe Iordanescu (Microsoft)
Hyperparameter optimization for machine leaning is complex, requires advanced optimization techniques, and can be implemented as a generic framework decoupled from specific details of algorithms. Fidan Boylu Uz, Mario Bourgoin, and George Iordanescu apply such a framework to tasks like object detection and text matching in a transparent, scalable, and easy-to-manage way in a cloud service. Read more.
Add to your personal schedule
1:45pm2:25pm Wednesday, March 18, 2020
Location: 210 F
Daniel Jeffries (Pachyderm)
With algorithms making more and more decisions in our lives, from who gets a job to who gets hired and fired, and even who goes to jail, it’s more critical than ever that we make AI auditable and explainable in the real world. Daniel Jeffries breaks down how you can make your AI and ML systems auditable and transparent right now with a few classic IT techniques your team already knows well. Read more.
Add to your personal schedule
2:35pm3:15pm Wednesday, March 18, 2020
Location: 210 E
Digital brands focus heavily on personalizing consumers' experience at every single touchpoint. In order to engage with consumers in the most relevant ways, Lily AI helps brands dissect and understand how their consumers interact with their products, more specifically with the product features. Sowmiya Chocka Narayanan explores the lessons learned building AI-powered personalization for fashion. Read more.
Add to your personal schedule
2:35pm3:15pm Wednesday, March 18, 2020
Location: 210 F
Moin Nadeem (Intel)
The real world is highly biased, but we still train AI models on that data. This leads to models that are highly offensive and discriminatory. For instance, models have learned that male engineers are preferable, and therefore discriminate when used in hiring. Moin Nadeem explores how to assess the social biases that popular models exhibit and how to leverage this to create a more fair model. Read more.
Add to your personal schedule
4:15pm4:55pm Wednesday, March 18, 2020
Location: 210 E
AI techniques are finding applications in a wide range of applications. Crowd-counting deep learning models have been used to count people, animals, and microscopic cells. Srikanth Gopalakrishnan introduces novel crowd-counting techniques and their applications, including a pharma case study to show how it was used for drug discovery to bring about 98% savings in drug characterization efforts. Read more.
Add to your personal schedule
4:15pm4:55pm Wednesday, March 18, 2020
Location: 210 B
Roshan Satish (DocuSign), Michael Chertushkin (John Snow Labs)
Roshan Satish and Michael Chertushkin lead you through a real-world case study about applying state-of-the-art deep learning techniques to a pipeline that combines computer vision (CV), optical character recognition (OCR), and natural language processing (NLP) at DocuSign. You'll discover how the project delivered on its extreme interpretability, scalability, and compliance requirements. Read more.
Add to your personal schedule
4:15pm4:55pm Wednesday, March 18, 2020
Location: 210 F
Luyang Wang (Restaurant Brands International), Jiao(Jennie) Wang (Intel)
Lu Wang and Jennie Wang explain how to build a real-time menu recommendation system to leverage attention networks using Spark, Analytics Zoo, and MXNet in the cloud. You'll learn how to deploy the model and serve the real-time recommendation using both cloud and on-device infrastructure on Burger King’s production environment. Read more.

Contact us

confreg@oreilly.com

For conference registration information and customer service

partners@oreilly.com

For more information on community discounts and trade opportunities with O’Reilly conferences

Become a sponsor

For information on exhibiting or sponsoring a conference

pr@oreilly.com

For media/analyst press inquires