Free digital event, hosted by NVIDIA.
November 8, 2023 9:30 a.m. PT
Join industry leaders, researchers, and Kaggle Grandmasters to learn how emerging data science tools and innovations can drive new insights from your data.
Data science is powering modern businesses, foundational to transformative applications like LLMs, recommender systems, fraud detection, cybersecurity, and more. During the summit, you’ll engage in interactive online discussions on the technologies that power these applications—like vector search, graph analytics, and large-scale ML—and how you can optimize them for your use case.
Join us for the AI and Data Science Virtual Summit opening keynote, where you’ll hear about the latest in accelerated computing for data science and machine learning.
NetworkX has established itself as the go-to library for data scientists using Python to study graphs and networks, and for obvious reasons: it's very easy to use, supports a wide variety of graph algorithms, and has a great community supporting it.
But as we amass more and more data, many graph problems reach sizes that exceed the practical capabilities of NetworkX's pure Python implementation, forcing users to abandon NetworkX for another library that's often harder to use.
In response, NetworkX has added the ability to seamlessly dispatch to optimized backends, giving users both ease of use and scalability. This presentation will cover NetworkX for graph analytics and how users can take advantage of dispatching to work with graph data in ways previously not possible.
Pandas is the go-to library in Python for working with tabular data, but it quickly becomes slow as dataset sizes grow into the gigabytes. cuDF's new pandas accelerator mode (cudf.pandas) solves this problem, bringing the speed of cuDF to every pandas workflow with zero code change required.
But we’re not stopping there. cuML, the scikit-learn-like GPU machine learning library, now provides a unified CPU/GPU interface that enables you accelerate your machine learning workflows and develop with a single, hardware agnostic codebase.
Join us for a captivating panel discussion where distinguished Kaggle Grandmasters will share their journeys in the world of data science competitions. Explore their personal stories of triumph over challenging competitions and discover the core techniques and strategies that propelled them to the pinnacle of Kaggle success. Aspiring data scientists will get practical tips for improving Kaggle rankings and expanding their data science skill sets.
Nearest neighbor vector search has seen substantial interest in recent years with its use in areas such as recommender and search systems, unsupervised learning, and new applications such as memory components for large language models (LLMs).
This talk will provide background on the vector search problem and the three main tradeoffs implicit in its execution: quality of results, data storage, and computation requirements. Meta's Faiss (Facebook AI Similarity Search) library, originating from 2016, provides an algorithmic toolkit enabling many of these tradeoffs which aren't often exposed by vector search libraries and services, or understood by their users. The history of the Faiss library and associated techniques will be detailed, including GPU-specific challenges for their realization. Interesting new applications of vector search such as approximate computing and data compression for machine learning will also be discussed.
Finally, we will introduce RAPIDS RAFT and how Faiss is leveraging RAFT to continue pushing forward the use of GPUs for vector similarity search.
This panel discussion will bring together experts with years of industry experience to discuss real-world applications of data science and machine learning technologies. Gain valuable perspectives on keeping pace with the rapidly evolving landscape, and discover invaluable advice for aspiring professionals aiming to lead data science and machine learning initiatives.
Founding Researcher | Fast.ai Fast.ai
Senior Principal Data Scientist | H2O.ai H2O.ai
All summit registrants will receive exclusive complimentary access to a new online, self-paced data science course from the NVIDIA Deep Learning Institute following the event.
Get an overview of RAPIDS, an open-source suite of GPU-accelerated libraries with APIs that mirror the most popular open-source data tools.
Dive deeper with our RAPIDS sessions from GTC 2023 all in one playlist.
Access eight different tutorials and cheat sheets introducing the RAPIDS ecosystem.
Sign up to receive the latest data science news from NVIDIA.
Send me the latest news, announcements, and more from NVIDIA about Enterprise Business Solutions and Developer Technology & Tools.
Send me the latest news, announcements, and more from NVIDIA about:
(Optional). You can unsubscribe at any time.
NVIDIA Privacy Policy