Presented By O'Reilly and Cloudera
Make Data Work
Feb 17–20, 2015 • San Jose, CA

Schedule: Hardcore Data Science sessions

Wednesday, February 18

9:00am–5:00pm Wednesday, 02/18/2015
Location: LL20 BC
Ben Lorica (O'Reilly Media), Ben Recht (University of California, Berkeley), Chris Re (Stanford University | Apple), Maya Gupta (Google), Alyosha Efros (UC Berkeley), Eamonn Keogh (University of California - Riverside), John Myles White (Facebook), Fei-Fei Li (Stanford University), Tara Sainath (Google), Michael Jordan (UC Berkeley), Anima Anandkumar (UC Irvine), John Canny (UC Berkeley), David Andrzejewski (Sumo Logic)
Average rating: ****.
(4.86, 7 ratings)
All-Day: Strata's regular data science track has great talks with real world experience from leading edge speakers. But we didn't just stop there—we added the Hardcore Data Science day to give you a chance to go even deeper. The Hardcore day will add new techniques and technologies to your data science toolbox, shared by leading data science practitioners from startups, industry, consulting... Read more.
9:05am–9:45am Wednesday, 02/18/2015
Location: LL20 BC.
Tara Sainath (Google)
Average rating: ****.
(4.50, 4 ratings)
DNNs were first explored for acoustic modeling, where numerous research labs demonstrated improvements in WER between 10-40% relative. In this talk, I will provide an overview of the latest improvements in deep learning across various research labs since the initial inception. Read more.
9:45am–10:30am Wednesday, 02/18/2015
Location: LL20 BC.
Michael Jordan (UC Berkeley)
Average rating: *****
(5.00, 3 ratings)
In this talk we show how statistical decision theory provides a mathematical point of departure for achieving such a blending. We develop theoretical tradeoffs between statistical risk, amount of data and "externalities" such as computation, communication and privacy. We develop procedures that allow one to choose desired operating points along such tradeoff curves. Read more.
10:30am–11:00am Wednesday, 02/18/2015
Location: LL20 BC.
Located on the Concourse level. Read more.
11:00am–11:30am Wednesday, 02/18/2015
Location: LL20 BC.
Maya Gupta (Google)
Average rating: ****.
(4.67, 3 ratings)
What makes a large machine learning system more interpretable and robust in practice? How do we take into account engineer's prior information about signals? We'll discuss the importance of monotonicity, smoothness, semantically-meaningful inputs and outputs, and designing algorithms that are easy to debug. Read more.
11:30am–12:00pm Wednesday, 02/18/2015
Location: LL20 BC.
Alyosha Efros (UC Berkeley)
Average rating: ****.
(4.88, 8 ratings)
In this talk, I will describe some of our efforts to bypass the "language bottleneck" and other information to help in visual understanding and visual data mining. Read more.
12:00pm–12:30pm Wednesday, 02/18/2015
Location: LL20 BC.
Eamonn Keogh (University of California - Riverside)
Average rating: ****.
(4.67, 6 ratings)
In this talk I will argue that, relative to other types of data (text, social networks etc), time series data is relatively underexploited, and that many opportunities are available for novel commercial applications and scientific discoveries. Read more.
12:30pm–1:30pm Wednesday, 02/18/2015
Location: LL20 BC.
Located in 230 A - 230 C. Read more.
1:30pm–2:00pm Wednesday, 02/18/2015
Location: LL20 BC.
Anima Anandkumar (UC Irvine)
Average rating: ****.
(4.00, 6 ratings)
I will demonstrate how to exploit tensor methods for learning. Tensors are higher order generalizations of matrices, and are useful for representing rich information structures. Tensor factorization involves finding a compact representation of the tensor using simple linear and multilinear algebra. Read more.
2:00pm–2:30pm Wednesday, 02/18/2015
Location: LL20 BC.
John Canny (UC Berkeley)
Average rating: ***..
(3.86, 7 ratings)
How fast can machine learning (ML) and graph algorithms be? In "roofline" design, every kernel is driven toward the limits imposed by CPU, memory, network etc. "Codesign" pairs efficient algorithms with complementary hardware. These methods can lead to dramatic improvements in single node performance: BIDMach is a toolkit for machine learning that uses rooflined design and GPUs to... Read more.
2:30pm–3:00pm Wednesday, 02/18/2015
Location: LL20 BC.
Fei-Fei Li (Stanford University)
Average rating: ****.
(4.14, 7 ratings)
In this talk, I will give an overview of what computer vision technology is about and its brief history. I will particularly emphasize on what we call the “three pillars” of AI in our quest for visual intelligence: data, learning and knowledge. Read more.
3:00pm–3:30pm Wednesday, 02/18/2015
Location: LL20 BC.
Located on the Concourse level. Read more.
3:30pm–4:00pm Wednesday, 02/18/2015
Location: LL20 BC.
David Andrzejewski (Sumo Logic)
Average rating: ***..
(3.60, 5 ratings)
Many of the millions of events logged inside a given software system are not isolated occurrences, but rather links in richly interconnected causal chains. However, classic SQL-style aggregation cannot easily capture this underlying structure. This talk discusses how graph mining techniques can surface high-value insights from the relationships between logged events. Read more.
4:00pm–4:30pm Wednesday, 02/18/2015
Location: LL20 BC.
John Myles White (Facebook)
Average rating: ***..
(3.67, 3 ratings)
In this talk, I'll describe the ways in which Julia improves upon the current generation of languages used for data science. Read more.
4:30pm–5:00pm Wednesday, 02/18/2015
Location: LL20 BC.
Chris Re (Stanford University | Apple)
Average rating: *****
(5.00, 3 ratings)
We describe how DeepDive is being used in a range of tasks from diagnosing rare diseases to drug purposing to filling out the tree of life. DeepDive helps to create knowledge bases that meet--and sometimes even exceed--human-level quality and to perform predictive analytics on top of this data. Read more.