Talk | MLOps & Data Engineering
Michelle is cofounder & CEO of noteable.io, an early-stage startup that’s building the first enterprise-grade platform for Jupyter notebooks. Before starting Noteable, she led the Big Data Tools engineering team at Netflix, where she was responsible for platform innovation and analytics tooling for Netflix’s industry-leading data platform. Prior to that, she led data engineering, data management, and platform architecture for GoDaddy, where she set a TPS record for SQL Server and helped pioneer Hadoop data warehousing techniques.
Talk | Cybersecurity
This session will examine the evolution of threat actor techniques, tactics and procedures, and take a look at where threats are evolving. It will cover, through use cases, basic strategies every incident response plan should contain and in particular, examine the roles situational awareness and information sharing play in helping organizations be better prepared to respond to incidents before, during and after they happen.
Denise Anderson, MBA, is President and CEO of the Health Information Sharing and Analysis Center (H-ISAC), a non-profit organization dedicated to protecting the global health sector from physical and cyber attacks and incidents through dissemination of trusted and timely information.
Denise currently serves as Chair of the National Council of ISACs, sits on the Board of Directors for the Global Resilience Federation (GRF) and the Executive Committee of the Cyber Working Group for the Health and Public Health Sector Coordinating Council. In addition she participates in numerous industry advisory groups and initiatives and has spoken at events all over the globe.
Denise was certified as an EMT (B), and Firefighter I/II and Instructor I/II in the state of Virginia for twenty years and was an Adjunct Instructor at the Fire and Rescue Academy in Fairfax County, Virginia for ten years.
She is a graduate of the Executive Leaders Program at the Naval Postgraduate School Center for Homeland Defense and Security.
Talk | Deep Learning | Intermediate
Tracking objects is a foundational task for video analysis. It is the engine for smart cities, autonomous driving, and building management. However, although many methods have been proposed in top-tier research conferences for the task each year, most of our production systems use half a decade old techniques. In this talk, I will explain the reason behind the gap in research and production, with intuition and experimental results. Further, I will introduce our recent works to address the issues. Our new methods can easily leverage large-scale datasets and learn to track objects in diverse scenarios. In the end, I will provide tools for industry practitioners to build trackers on their data without diving into complicated parameter tuning and expensive optimization. You will learn to make robust, simple, and performant tracking modules to supercharge your video analysis engines.
Fisher Yu is an Assistant Professor at ETH Zürich in Switzerland. He obtained his Ph.D. degree from Princeton University and became a postdoctoral researcher at UC Berkeley. He is now leading the Visual Intelligence and Systems (VIS) group at ETH Zürich. His goal is to build perceptual systems capable of performing complex tasks in complex environments. His research is at the junction of machine learning, computer vision and robotics. He currently works on closing the loop between vision and action. His works on image representation learning and large-scale datasets, especially dilated convolutions and the BDD100K dataset, have become essential parts of computer vision research. More info is available at https://www.yf.io
Talk | Machine Learning | Intermediate-Advanced
Adriana Romero Soriano is a research scientist at Facebook AI Research and an adjunct professor at McGill University. Her research focuses on developing models and algorithms that are able to learn from multi-modal data, reason about conceptual relations, and leverage active acquisition strategies to mitigate their uncertainties. The playground of her research has been defined by problems which require inferring full observations from limited sensory data. She completed her postdoctoral studies at Mila, where she was advised by Prof. Yoshua Bengio. Her postdoctoral research revolved around deep learning techniques to tackle biomedical challenges, such as the ones posed by multi-modal data, high dimensional data and graph structured data. She received her Ph.D. from University of Barcelona in 2015 with a thesis on assisting the training of deep neural networks, advised by Dr. Carlo Gatta.
Talk | Research Frontiers | Beginner-Intermediate
In this talk, we will learn about Pyro (http://pyro.ai) a PPL built on PyTorch. We will discuss what probabilistic programming is, and how we can integrate it with deep learning to tackle open machine learning problems in generative modeling. We will talk about approximate inference techniques such as variational inference, and walk through some of the tools and examples to make inference on models automatic. If you are a data scientist, an ML engineer, or an ML researcher, this talk will be of interest to you!
JP is a research scientist at Facebook where he works on probabilistic programming, approximate inference, and Bayesian nonparametrics. He is a founding coauthor of the probabilistic programming language Pyro. The main question that guides his research is: how do we build and perform inference on models in an automatic yet principled way? Prior to Facebook, he was at Uber AI Labs working at the intersection of deep learning and statistics, focusing on time series forecasting and mapping for self driving cars.
Tutorial | Machine Learning
Talk | Deep Learning
James Le currently runs Data Relations at Superb AI, a Series A ML data management startup. As part of his role, James executes content and partnership initiatives – while working cross-functionally with growth, product, customer success, sales, marketing, and community functions to drive Go-To-Market strategy.
Before joining Superb AI, he completed his Computer Science Master’s degree at RIT, where his research thesis lies at the intersection of deep learning and recommendation systems. Outside of work, he is highly active in the broader data and ML community – writing data-centric blog posts, hosting a data-focused podcast, and teaching an online course for ML practitioners.
Talk | Machine Learning | Responsible AI | Intermediate
Individual Fairness (IF) is a very intuitive and desirable notion of fairness: we want ML models to treat similar individuals similarly, that is, to be fair for every person. For example, two resumes of individuals that only differ in their name and gender pronouns should be treated similarly by the model. Despite the intuition, training ML/AI models that abide by this rule in theory and in practice poses several challenges. In this talk, I will introduce a notion of Distributional Individual Fairness (DIF) highlighting similarities and differences with the original notion of IF introduced by Dwork et al. in 2011. DIF suggests a transport-based regularizer that is easy to incorporate into modern training algorithms while controlling the fairness-accuracy tradeoff by varying the regularization strength. Corresponding algorithm guarantees to train certifiably fair ML models theoretically and achieves individual fairness in practice on a variety of tasks. DIF can also be readily extended to other ML problems, such as Learning to Rank.
Mikhail is a Research Staff Member at IBM Research and MIT-IBM Watson AI Lab in Cambridge, Massachusetts. His research interests are Model fusion and federated learning; Algorithmic fairness; Applications of optimal transport in machine learning; Bayesian (nonparametric) modeling and inference. Before joining IBM, he completed Ph.D. in Statistics at the University of Michigan, where he worked with Long Nguyen. He received his bachelor’s degree in applied mathematics and physics from the Moscow Institute of Physics and Technology.
Talk | Responsible AI | Beginner
How can artificial intelligence and open data science tackle the twin challenges of climate change and reversing biodiversity loss? Come hear about some of the successes in addressing these challenges and get some tips on how you can help. Topics include case studies on how the World Wildlife Fund and others are using AI to help predict deforestation, monitor the health of protected areas, and map carbon stocks, going into some detail about the artificial intelligence techniques applied. The talk will also address difficulties involved with AI applications due to the sensitive nature of much of the relevant data. Pointers for further exploration will be given throughout the talk, and it will close with some open questions you can help answer.
Dave Thau is WWF’s Data and Technology Global Lead Scientist with him over 30 years of software development and conservation experience. He is also a member of the IPBES Knowledge and Data taskforce. Prior to WWF, Dave worked at the California Academy of Sciences, the Kansas University Museum of Natural History, and Google where he helped launch Google Earth Engine.
Dave’s work focuses on the fields of data management, sustainability, artificial intelligence, and remote sensing. He holds degrees from the University of California, Los Angeles, the University of Michigan, Ann Arbor, and a doctorate in computer science from the University of California, Davis. He also has an ant named in his honor – the charming Plectroctena thaui.
Talk | Deep Learning | Machine Learning | Intermediate
Machine learning has traditionally relied on creating models around data that can be represented in tabular format such as SQL tables, Pandas dataframes, and the like. Inherent in this data is the assumption that there is no relationship between each entry (row) of the data. In certain cases this is an accurate assumption. However, there are many common use cases for machine learning where this assumption is not entirely accurate. In these cases, by considering the relationships among those individual data points, models can be significantly enhanced and measurable improvements can be made to the appropriate metrics of that model. Such use cases can include common data science and machine learning tasks such as churn prediction and automated recommendation engines.
In this talk we will compare and contrast models created with individual data points to those made entirely with graphs and hybrids of the two. We will explore a variety of techniques that are used for creating graph embeddings, the vectors for representing graphs that are created in a similar fashion to the feature engineering and vector embeddings associated with traditional machine learning. We will focus on the optimization of the graph embeddings and explore some real-world examples of their use individually and in conjunction with the traditional types of machine learning embeddings. Special emphasis will be placed on the benefits of using graph embeddings with significant class imbalance. We will also discuss the use of these embeddings with traditional machine learning packages and workflows, such as through the use of scikit-learn and TensorFlow.
Dr. Clair Sullivan is currently a graph data science advocate at Neo4j, working to expand the community of data scientists and machine learning engineers using graphs to solve challenging problems. She received her doctorate degree in nuclear engineering from the University of Michigan in 2002. After that, she began her career in nuclear emergency response at Los Alamos National Laboratory where her research involved signal processing of spectroscopic data. She spent 4 years working in the federal government on related subjects and returned to academic research in 2012 as an assistant professor in the Department of Nuclear, Plasma, and Radiological Engineering at the University of Illinois at Urbana-Champaign. While there, her research focused on using machine learning to analyze the data from large sensor networks. Deciding to focus more on machine learning, she accepted a job at GitHub as a machine learning engineer while maintaining adjunct assistant professor status at the University of Illinois. Additionally, she founded a company, La Neige Analytics, whose purpose is to provide data science expertise to the ski industry. She has authored 4 book chapters, over 20 peer-reviewed papers, and more than 30 conference papers. Dr. Sullivan was the recipient of the DARPA Young Faculty Award in 2014 and the American Nuclear Society’s Mary J. Oestmann Professional Women’s Achievement Award in 2015.
Talk | Machine Learning | Research Frontiers | Intermediate
In the last decade many different types of neural networks have been developed. They showed us the amazing power and opportunities of machine learning. Everywhere in the world processes are replaced by ML algorithms, people are matched with their dream job, products are recommended and cars are driven automatically. It is truly amazing what we can do with such models. On the other hand, when you take a critical look, the whole training process is not that efficient. We have to feed models with millions of labeled images or text inputs to make sure your algorithm will perform well. And thinks of what happens in this training process. Each input goes through many layers where multiplications and ReLu or sigmoid functions are applied to each item from the input. Forward and backwards! Due to back propagation. Of course, with all the available computer in the form of GPU’s this is not really an issue. However this cost a lot of energy. With that in mind we do know that neural networks are sort of based on the way humans learn. Except that the human brain is much more energy efficient. Could we achieve that same energy-efficient level in artificial neural networks? The answer is yes!
In this talk I will show you what is often called the third generation neural networks: Spiking Neural Networks. Based on the biological processes in the brain this kind of neural network uses discrete spikes and sparse communication to learn. I will give short introduction in some biological processes in the human brain and from there we will define spiking neural networks. We will discuss the downsides compared to artificial neural networks due to their discontinuous nature and I will show a resolution to that. The accuracy maybe still falls short of the artificial neural networks but the field is evolving and I will show the great potential of these networks. You will also get an overview of some existing frameworks based on Pytorch.
Let’s go for great accuracy and major energy savings
As a graduated Mathematician I’m particularly interested in the techniques and math behind algorithms. How do they search for the optimal solution and why is one algorithm faster than the other? In my work as a Data Scientist I develop algorithms or adapt existing solutions to customer needs and put them into production such they can get the most value out of it. In my own time I love to read popular scientific articles or books about mathematics, physics or astrophysics. Besides this I love traveling and cycling.
Talk | Machine Learning Data Engineering and MLOps | Intermediate
At YooMoney, we use ML Models extensively for different tasks from Anti-fraud to NLP.
We started with a Data Scientist who used jupyter and then copy-pasted model code to flask vs pickle and zipped it for production. But it was a labor-intensive and hardly sclable process, so we begin to introduce MLOps.
My talk will cover MLOps practices—a way to streamline the model development process and automate it as much as possible. In general, at least some of them are an attempt to use Software Development practices in Machine Learning experimentation and production.
From one point of view, this task was relatively easy for our company: we already have CI/CD in place for regular applications, so why not just use them for ML purposes?
But when it comes to implementation, one might understand that it is not such a straightforward process. I will go through the main stages of the MLOps pipeline, explaining the challenges and solutions to overcome them.
The first stage is Model Development. On the one hand, it looks like regular software development (writing some code), but on the other hand, it doesn’t as it requires access to a lot of datasets and DWHs (preferably, with live data – in case of Fintech, as well as Medical data, it might be challenging), so we have to solve Security issues like introducing IDM interfaces for ‘sets of datasets””.
There are some issues with code writing tools on this stage: Data Scientists often write code in jupyter notebooks instead of IDEs like Idea/Eclipse/VS, which is not directly suitable for creating a standalone application and requires some additional effort on the merge phase during commit. We manage to solve it with the jupytext module, it helps sync both ways between py and ipynb, storing py-files as the main reference in git.
The next stage is Preparing for Production. It starts with Model Risk Evaluation: we will briefly mention the probability-impact matrix, and name the risks that can be mitigated using MLOps like operational or data drift.
QA (testing practices suitable for ML) also plays an important role in the MLOps Process. As we aim to automate model’s lifecycle, short testing period will definitely help achieve this aim. The cheapest and most straightforward solution here is to use a tool like pytest, but it only works until other platforms like Scala applications are introduced, so alternatives like Kotlin autotests should be considered. As for the testing strategy, a few solutions can be used here (testing on a reference dataset, accessing ground truth, checking business metrics, and so on).
After that, we proceed to building the environment for real-time inference. As OReilly’s “Introduction to MLOps” suggests, this part of the process should come as early as possible, before the model is prepared (or even before model development starts). With basic modules like scikit it is relatively simple process, but when it comes to Tensorflow, for example – situation changes (who have tried just run latest version of tensorflow with latest version of python and all the libraries? How often it works with a first attempt?). So here we have to solve platform-specific problems (like it is better to use virtualenv or docker for dependencies) and more general ones: what kind of tools will be used for inference platform management. Should it be a single machine, Kubernetes cluster, or whatever? Our current solution is horizontal scaling with a balancer, and we’re aiming to use docker under Kubernetes as the target platform.
Deploying also has a lot of things to consider. We start with building artifacts for deployment. What should come with a model, just a serialized object? Or a reproducible research set including Data? We started with pickle / hdf5, and they suited perfectly until Scala models were introduced. Now we have to switch to another technology choosing different formats such as PMML, PFA, ONNX, or POJO (main pros and cons will be discussed during the presentation).
Release (which is not the same as deployment). At this stage, we start using the new model. Some Risks discussed above might be mitigated here: for example, the operational one can be dealt with using Canary releases or blue-green releases.
Last but not least, we might combine a few of the above discussed techniques in Monitoring. Here we need a Model Repository or some other way of detecting the model version (writing this version to a log works well too). Also, testing practices can be used to ensure that the model performs well in the production environment (from heartbeat/ping request till human evaluation for a portion of requests).
The main idea for MLOps is something similar to DevOps: identify labor-intensive parts of the Model lifecycle, choose the ones that can be automated and match appropriate tools for these parts (example was given above). This approach makes Model development more predictable and ensures that highly-qualified people like Data Scientists or Subject Matter Experts can focus on their specific task, leaving almost all infrastructure-related tasks to automated tools.
Evgenii is the Head of Data Engineering and Data Science team at YooMoney, the leading payment service provider on the CIS market. Evgeny and his team have completed a wide range of projects including an accounting system based on blockchain technologies (as an analyst), a BRE+ML-based antifraud engine (as an architect and project manager), Business Intelligence solutions (as a developer, analyst, architect, project manager), and many others. Currently, Evgenii participates in ML projects as ML Architect and Project Manager.
Talk | MLOps & Data Engineering | Intermediate
Development tools such as Jupyter are prevalent among data scientists because they provide an environment to explore data visually and interactively. However, when deploying a project, we must ensure the analysis can run reliably in a production environment like Airflow or Argo; this causes data scientists to move code back and forth between their notebooks and these production tools. Furthermore, data scientists have to learn an unfamiliar framework and write pipeline code, which severely delays the deployment process.
Ploomber solves this problem by providing:
1. A workflow orchestrator that automatically infers task execution order using static analysis.
2. A sensible layout to bootstrap projects.
3. A development environment integrated with Jupyter.
4. Capabilities to export to production systems (Airflow and Argo) without code changes.
This talk develops and deploys a Machine Learning pipeline in 30 minutes to demonstrate how Ploomber streamlines the Machine Learning development and deployment process.
Who and why
This talk is for data scientists (with experience developing Machine Learning projects) looking to enhance their workflow. Experience with production tools such as Airflow or Argo is not necessary.
The talk has two objectives:
1. Advocate for more development-friendly tools that let data scientists focus on analyzing data and taking off popular production tools’ overhead.
2. Demonstrate an example workflow using Ploomber where a pipeline is developed interactively (using Jupyter) and deployed without code changes.
Eduardo is interested in developing tools to deliver reliable Machine Learning products. Towards that end, he created Ploomber, an open-source Python library to compose production-ready data workflows. Eduardo holds an M.S in Data Science from Columbia University, where he took part in Computational Neuroscience research. Eduardo started his Data Science career in 2015 at the Center for Data Science and Public Policy at The University of Chicago.
Talk | Machine Learning | Intermediate
Survival models describe how long it will take for some important event to occur. Because they account for censored data and avoid arbitrary binarization thresholds, they are a natural fit for many applications. In churn prediction, for example, it can be more useful to model the time until a subscriber churns, instead of the probability of churn over some arbitrary time window. Similarly, for hardware engineering, it can be more valuable to know the likely time until a piece fails instead of the probability of failure over a fixed time frame.
Despite the name survival, not all applications involve negative outcome events; in some cases, we want to reduce the time-to-event. For example, we can use survival models to decide between competing sales strategies; the better policy is the one with the shortest time-to-conversion.
While conceptually elegant, survival analysis has historically been popular only within a handful of applied domains, especially clinical research. The machine learning community has recently taken notice, however, and survival analysis is gaining traction within research, applications, and software circles.
Our goal in this talk is to help the audience add survival modeling to their working data science tool belt. We’ll first introduce basic concepts of survival analysis like censored data, duration matrices, and survival curves. We’ll then show when to consider using survival models instead of other methods, how to use popular Python survival analysis tools Lifelines, Scikit-survival, and Convoys, and how to interpret model results for either prediction or decision-making. Throughout, we’ll emphasize the machine learning perspectives on the topic.
Brian Kent is the founder of The Crosstab Kite, a publication for professional data scientists solving real-world challenges. He writes about survival analysis, data-driven decision-making, data science tools, and big picture trends in statistical modeling.
Prior to The Crosstab Kite, Brian worked in the FinTech space as Director of Data Science & Machine Learning at Credit Sesame. Before that, he was a machine learning engineer at Apple, where he worked on autonomous systems,
personalized health, and silicon engineering.
Talk | Machine Learning | Beginner-Intermediate
It is important to efficiently determine the health of large complex systems by detecting anomalous behavior, where anomalies in the system data can help detect if there is a failure or an impending failure. The goal is to detect anomalous behavior before it escalates to severe service degradation or a service impacting outage.
In this talk, using sequential multivariate system performance data, we present the application of multivariate change detection algorithms and visual analytics methods for detecting and diagnosing anomalous behavior with low latency in a large networking system. A brief overview of anomaly detection concepts will also be presented.
Multivariate change detection algorithms based on non-parametric change detection methods are applied to the data to detect anomalies and present diagnostic information at fine time granularity. We identify whether a change point is a single time stamp (pointwise anomaly) or a collection of time stamps (collective anomaly) that does not conform with the general pattern of data.
Two unsupervised change point detection methods are used, namely, the Bayesian approach and the distance-based approach. For the Bayesian approach, we deploy the following R packages: changepoint.mv and anomaly. The R package ecp is selected for the distance-based change detection approach. An advantage of the changepoint.mv package is that it also provides diagnostic capability in terms of explicitly identifying both the change point location and the variables associated with the change point.
The R packages used for change detection will be described in terms of their capabilities and characteristics, and the R code used for the analysis will be shared. In addition, the use of self-organizing maps (using the R kohonen package) for visual analytics will be presented. We demonstrate our methods with real data.
Veena is a network analytics and reliability consultant to business groups. Former research lead at Nokia Bell Labs where she led network reliability and analytics research. Other expertise includes software and data reliability, data visualization, network resiliency.
Talk | Responsible AI | Beginner-Intermediate
AI has made amazing technological advances possible; as the field matures, the question for AI practitioners has shifted from “can we do it?” to “should we do it?”. In this talk, Dr. Tempest van Schaik will share her Responsible AI (RAI) journey, from ethical concerns in AI projects, to turning high-level RAI principles into code, and the foundation of an RAI review board that oversees projects for the team. She will share some of the practical RAI tools and techniques that can be used throughout the AI lifecycle, special RAI considerations for healthcare, and the experts she looks to as she continues in this journey.
Talk | Deep Learning | Machine Learning | Beginner-Intermediate
Extracting key-fields from a variety of document types remains a challenging problem. Services such as AWS, Google Cloud and open-source alternatives provide text extraction to “digitize” images or pdfs, returning phrases, words and characters. Processing these outputs is unscalable and error-prone as varied documents require different heuristics, rules or models and new types are uploaded daily. In addition, a performance ceiling exists as downstream models rely on good yet imperfect OCR algorithms upstream.
We propose an end-to-end solution utilizing image-based deep learning to automatically extract important text-fields from documents of various templates and sources. Computer vision algorithms utilizing deep learning produce state-of-the-art classification accuracy and generalizability through training on millions of images. We compare the in-house model accuracy, processing time and cost with 3rd party services and found favorable results to automatically extract important fields from documents.
Bill.com is working to build a paperless future. We process millions of documents a year ranging from invoices, contracts, receipts and a variety of others. Understanding those documents is critical to building intelligent products for our users.
I’m the Chief Data Scientist at Bill.com and have many years of experience as a scientist and researcher. My recent focus is in machine learning, deep learning, applied statistics and engineering. Before, I was a Postdoctoral Scholar at Lawrence Berkeley National Lab, received my PhD in Physics from Boston University and my B.S. in Astrophysics from University of California Santa Cruz. I have 2 patents and 11 publications to date and have spoken about data at various conferences around the world.
Recruitment and hiring has inherent bias that hinders companies from hiring the candidates who are best fit for the job. This hurts not only individual career paths and diversity efforts, but also the growth and success of organizations. By leveraging AI and career data from over 1 billion people, Ashutosh helped build a system free of bias that is shifting how companies recruit, hire, retain, and grow talent. Equal parity algorithms and audit and monitoring processes create a transparent system that is independent of race, gender, ethnicity, age, and other characteristics. Ashutosh will go into detail about how AI removes fairness concerns by increasing transparency and mitigating risk for unconscious bias. He’ll delve into how this system is helping remap the traditional career journey, increasing diversity at every level and helping Americans get back to work post-pandemic. Finally, Ashutosh will share case studies of how HR departments at leading organizations are using these technologies to make hiring more equitable and inclusive.
Business Talk | AI for Healthcare
In the upcoming stages of the Fourth Industrial Revolution, we are going to experience a paradigm shift in how we use Artificial Intelligence (AI) and Robotics to improve processes and enhance healthcare. During her presentation, Alishba will discuss various applications of AI, efforts to develop Artificial General Intelligence (AGI), soft robots, and how these technologies can be used to facilitate and enhance mental health practices, improve prosthetics and rehabilitation devices for recovering stroke patients. She will demonstrate this by walking through her work with San Jose State University where she used 3D printing and AI to develop a cheaper prosthetic that costs $700 vs the current price of $10k. As well, she will be providing insights through highlighting her work with Hanson Robotics on Sophia the Robot to improve manipulation techniques for robots and develop the next wave of humanoid robots with human-like intelligence. She will also highlight her research and research by labs such as the Harvard Biodesign Lab to use soft robotics and machine learning for the development of low-cost and easier to use portable rehabilitation gloves for recovering stroke patients. As well, Alishba will speak on her work with Kindred.Ai using imitation learning and telerobotics to develop more intelligent and safe human-robot interactions that can be used in medical and manufacturing settings. This interactive session will highlight current cutting-edge research being done in robotics and AI while diving deep into the concepts powering them and how these are being applied to key problem areas in medical and rehabilitation industries.
Alishba is an 18-year-old machine learning and blockchain developer who was named a Young Innovator to Watch at CES in 2020. At 15, she developed Honestblocks, a blockchain platform to track medication and put an end to counterfeit medication in supply chain systems for 2 million people in rural India. Part of this platform was eventually integrated into IBM Blockchain and is being used in various supply chain applications. She has applied her skills as an intern at various startups and companies such as TD Bank, Pngme, and Vestergaard. At TD Bank, she developed a new blockchain product to securely allow 2M+ clients to store their personal, financial data and access different financial services. She has worked as an ML Developer at Hanson Robotics to develop Neuro-Symbolic AI, RL, and Generative Grasping CNN approaches for Sophia the Robot. Alishba further applied this work with a master’s student and professor at San Jose State University, with support from the BLINC Lab, to reduce cost of prosthetics from $10k to $700, and make grasping more efficient.
Talk | Deep Learning | Research Frontiers | Intermediate
The ultimate goal of generative learning is to learn a model of how data is generated in the real world. Such models allow us to generate novel samples from a data distribution and have applications in (i) content generation (e.g., image, video, speech, text, music, and molecule generation) and (ii) representation learning and semi-supervised training (e.g., learning from limited labeled data). In this talk, I will first briefly review different classes of deep generative models, and then, I will focus on the recent developments in this field including the state-of-the-art variational auto-encoders (VAEs), denoising diffusion models, and generative adversarial networks (GANs). This talk will mostly focus on generative models designed for image synthesis, however, the techniques discussed in the talk can be easily applied to other data types.
Arash Vahdat is a senior research scientist at NVIDIA research specializing in machine learning and computer vision. Before joining NVIDIA, he was a research scientist at D-Wave Systems where he worked on deep generative learning and weakly supervised learning. Prior to D-Wave, Arash was a research faculty member at Simon Fraser University (SFU), where he led research on deep video analysis and taught graduate-level courses on big data analysis. Arash obtained his Ph.D. and MSc from SFU under Greg Mori’s supervision working on latent variable frameworks for visual analysis. His current areas of research include deep generative learning, weakly supervised learning, efficient neural networks, and probabilistic deep learning.
For every business, and particularly for a growing company like Wix, it is crucial to have full grasp over the future incoming cash flow. This allows us to optimally plan and allocate money for operation costs and investments. Being a public company, the forecasts are also important as part of guidance given to investors for the upcoming fiscal quarter or year.
Wix makes most of its revenue from paid subscriptions of new and existing users. The number of premium subscriptions and the amount of cash collected are the targets we want to forecast. In contrast to the usual approach of treating this as a time-series problem where we target specific dates, we can treat this as a regression problem where target values are defined by user registration date and age. This is what we call a cohort-based model.
Users that registered on a certain date represent a cohort. This cohort has its own features (e.g. size, country) which are joined into a table based on their registration date. In the dimension of time, regardless of age, all cohorts are subject to the same events, like seasonality, holidays, general trend, discounts, price changes. These features are joined to a table by the upgrade date, which defines the cohort age. Thus a row in a table should read something like: “cohort of users registered on 2021-01-01 that is 3 days old on upgrade date 2021-01-03 produced 100 subscriptions”.
This approach allows the use of regression models which normally can’t be applied to time-series data, for example GLM, GAM and GBM with Poisson or Tweedie distributions. In terms of error rates, these cohort-based models proved to be at par or better than time-series models like Prophet.
Nicolai Vicol is a Data Scientist at Wix, where he specializes in forecasting of new users, paid subscriptions, cash flows and generally everything related to time-series. He started his career as a quant in an investment bank, then switched to data science and IT, accumulating in total 9 years of experience in the field. Areas of interest: time series and forecasting, but also recommendation systems, search systems and operation research.
Talk | Deep Learning | Responsible AI | All Levels
This talk provides an introduction of how AI can be used to assist and inspire the artists and designers in their creative space. You will learn about how to use advanced features of TensorFlow 2.x, Keras and TensorFlow Lite to train and deploy models to applications. This talk is for data scientists and ML engineers who are interested in applying deep learning models to create generative art & design.
Generative art and design
TensorFlow 2.x and Keras
TensorFlow Lite and on-device ML
Margaret is an ML research engineer working on applying AI/ML to real-world applications from climate change to art and design. She writes and speaks at conferences about deep learning, computer vision, TensorFlow and on-device ML. She leads Google Developer Group (GDG) Seattle and Seattle Data/Analytics/Machine Learning. She is recognized for her expertise as a Google Developer Expert (GDE) for ML. She is also an avid artist who creates traditional, digital, and AI art – check out her artwork at margaretmz.art.
Talk | Deep Learning | Computer Vision | All Levels
Is one eye all you need? Can we learn robot perception from raw videos only? Can we get robust 3D depth estimation from a single monocular camera? In this talk, we will discuss recent research progress we made at TRI on self-supervised learning for 3D vision, its uses, limitations, and promising future directions combining self-supervision with other scalable sources of supervision like simulation.
Adrien Gaidon is the Head of Machine Learning Research at the Toyota Research Institute (TRI) in Los Altos, CA, USA. Adrien’s research focuses on scaling up ML for robot autonomy, spanning Scene and Behavior Understanding, Simulation for Deep Learning, 3D Computer Vision, and Self-Supervised Learning. He received his PhD from Microsoft Research – Inria Paris in 2012, has over 50 publications and patents in ML & Computer Vision (cf. Google Scholar), and his research is used in a variety of domains, including automated driving. You can find him at adriengaidon.com, on linkedin, and Twitter @adnothing.
Tutorial | NLP
Kumaran Ponnambalam is an AI and Big Data leader with 15+ years of experience. He is currently the Director of AI for Webex Contact Center at Cisco. He focuses on creating robust, scalable AI platforms and models to drive effective customer engagements. In his current and previous roles, he has built data pipelines, ML models, analytics and integrations around customer engagement. He has also authored several courses on the LinkedIn Learning Platform in Machine Learning and Big Data areas. He holds a MS in Information Technology and advanced certificates in Deep Learning and Data Science.
Workshop | MLOps & Data Engineering
Daniel Imberman is a PMC of the Apache Airflow project, core contributor of the Kubernetes Executor, and Strategy Engineer at Astronomer.io. He recieved a BS/MS at UC Santa Barbara with a focus in Distributed Systems and Machine Learning and is highly passionate about building the next generation of ML tooling.
Half-Day Training | Data Analytics
Matt is currently Growth Manager at Roboflow, a YCombinator-backed startup focusing on responsibly democratizing computer vision. Matt is also Managing Partner + Principal Data Scientist at BetaVector. His full-time professional data work spans finance, education, consumer-packaged goods, and politics and he earned General Assembly’s 2019 “Distinguished Faculty Member of the Year” award. Matt earned his Master’s degree in statistics from Ohio State. Matt is passionate about mentoring folx in data and tech careers. Matt also volunteers with Statistics Without Borders and currently serves on their Executive Committee as the Marketing & Communications Director.
Workshop | Deep Learning | Beginner-Intermediate
In this 90-minute workshop we will describe how to apply Deep Learning (DL) to a few Natural Language Processing (NLP) tasks. It is an introductory workshop, where the target audience is expected to have basic knowledge of neural networks but little or no NLP experience. The format is a mix of presenting theory and walking through programming examples.
You will learn about language models and word embeddings and get an idea of why they work. You will see examples of how to implement a seemingly complicated task like natural language translation using 300 lines of code, both in TensorFlow and PyTorch. The overall goal of this workshop is to connect the dots between basic neural networks and more complicated NLP architectures like BERT and GPT.
Magnus Ekman is a Director of Architecture at NVIDIA, where he leads an engineering team working on CPU performance and power efficiency. As the deep learning (DL) field exploded in the past few years, fueled by NVIDIA’s GPU technology and CUDA, he found himself in the midst of a company expanding beyond computer graphics and becoming a DL powerhouse. As a part of that journey, he challenged himself to stay up to date with the most recent developments in the field. In collaboration with NVIDIA Deep Learning Institute (DLI) he recently published the book “Learning Deep Learning: Theory and Practice of Neural Networks, Computer Vision, Natural Language Processing, and Transformers Using TensorFlow.”
Workshop | NLP
Comprehending natural language text with its first-hand challenges of ambiguity, synonymity and co-reference has been a long-standing problem in Natural Language Processing. The domain of Natural Language Processing has seen a tremendous amount of research and innovation in the past couple of years to tackle this problem and to implement high quality machine learning and AI solutions using natural text by abstracting the underlying workings of the algorithms. This essentially allowed for quick application of pre-trained models and integrated them in the real-world industry use-cases. Question-Answering is one such area that is crucial in all sectors like finance, media, chatbots to explore large text datasets and find insights quickly. You can either build a closed domain QA system for specific use-case or work with open domain systems using some of the open-sourced language models that have been pre-trained on terabytes of data on general knowledge base. Fine-tuning it based on the problem at hand to add additional information is the way to efficiently implement a machine learning solution. The general idea is to identify K relevant sentences from the training corpus for a question query, that will then find the span of text from sentences which answers the question.
This talk will highlight the general concepts and ways of implementing language model DistilBERT and fine tuning the base model to build an efficient question-answering model. This also ensures that using the available open source platforms we are able to have better business outputs as well as better environment because training a single AI model contributes to 5 cars’ lifetime worth of carbon emissions? Basic understanding of python is desirable. Code can be made available via GitHub for everyone to examine after the talk.
Jayeeta is a Data Scientist with 5+ years of industry experience. She recently led six-week NLP workshops in association with Women Who Code, Data Science track. Jayeeta has also been a speaker at International Conference on Machine Learning (ICML 2020), MLConf EU, WomenTech Global Conference, and Data Summit Connect. She works extensively on NLP projects where she gets to explore a lot of state-of-the-art models and build cool products and firmly believes that data is the best storyteller. Recently, Jayeeta joined MediaMath, a leader in the programmatic AdTech domain. Prior to this, she worked at Indellient, Omnicom, Deloitte, and Volvo Group. Jayeeta is also engaged with some amazing organizations to promote and inspire more women to take up STEM. Jayeeta received her Master of Science in Quantitative Methods and Modeling from City University of New York, NY, and Bachelor of Science in Economics and Statistics from West Bengal State University, India.
Website – https://jayeetap.wixsite.com/helloworld
Workshop | Deep Learning | Big Data Analytics | Beginner-Intermediate
Predicting the future has always been a fascinating topic. Now we have AI tools and techniques that can help us do it better than ever before. In this session, we’ll cover the fundamentals of solving time-series problems with AI, and show how it can be done with popular data science tools such as Pandas, TensorFlow, and the Google Cloud AI Platform.
We’ll start with how to visualize, transform, and split time-series data for use in an ML model. We’ll also discuss both statistical and machine learning techniques for predictive analytics. Finally, we’ll show how to train a demand forecasting model in the cloud and make predictions with it. Attendees can access Jupyter notebooks after the session to review the material in more detail.
Karl Weinmeister is a Cloud AI Advocacy Manager at Google, where he leads a team of data science experts who develop content and engage with communities worldwide. Karl has worked extensively in machine learning and cloud technologies. He was a contributor to one of the first AI-based crossword puzzle solvers that is still referenced today.
Workshop | Deep Learning | Machine Learning | Intermediate
Applications for audio based machine learning include virtual assistants, automatic speech recognition, speech to text, firearm locators, vehicle accident early detection, wildlife monitoring, audio anomaly detection, denoising, and music classification.
After completing this workshop, you will be able to use a short-time Fourier transform to convert audio into features suitable for use in machine learning models, and apply these features in a sound classification and sound detection task. You will be able to develop your own feature generation pipeline for audio data, and be able to implement and adapt published sound detection and classification models for your particular use case.
In part 1, we will discuss characteristics of sound waveforms, import sample audio, and produce spectrograms using a Short Time Fourier Transform (STFT). We will also investigate the consequence of different choices of rectangular, triangular, and Hann window functions used in STFT.
In part 2, we will work through a wildlife monitoring use case which will require using STFT to transform audio recordings of rainforest sounds into spectrograms, create time slices of these spectrograms, and classify the sound slices according to species. We will then extend the classification task to a sound detection class and create bounding boxes around time periods during which species calls are present.
This workshop will make use of a Jupyter Lab running inside a Docker container preloaded with required packages – Tensorflow and Librosa. Some familiarity with tensorflow and/or audio data will help with understanding, but is not required for this workshop.
After feature generation, the neural network training shares some similarity with image tasks, so this workshop may also be informative for those seeking to learn more about image classification and object detection.
Ryan Kasichainula is a data science instructor at Galvanize, Inc, an industry leader in technology education, with data science and software engineering immersive bootcamps. They are also an independent data consultant with experience in the technology, agriculture, energy, and pharmaceutical industries. Ryan enjoys applying data science techniques to a wide variety of domains, and they always have at least one side project in the works, usually in the realm of natural language generation.
Ron Li is a data science instructor and senior data scientist at Galvanize, Inc. Before that, He worked on machine learning and knowledge graphs at the Information Sciences Institute. Ron has published a 4.5-star rating book Essential Statistics for Non-STEM Data Analysts. He has also authored/co-authored several academic papers, taught data science to non-STEM professionals as pro bono service, and gave talks at conferences like PyData.
Half-Day Training | Deep Learning | Intermediate
GANs are one of the most useful DL techniques in recent years, particularly for the tasks of data synthesis. Estimating an intrinsic data distribution from a given dataset and generating newer data that look like one from the given dataset is one of the significant successes of GANs. In this talk, I will present the fundamental principle behind the GANs, a mathematical formulation of it, how to build GANs using TensorFlow APIs, and how to train them to generate new images. I will also cover some of the extended GANs architecture and provide a future roadmap for beginners and advanced users of the GANs framework. This talk will have two parts – a theory class followed by a half-day of tutorials where I will show baseline code to build GANs models and train them; it will be followed by inference of image synthesis.
Ajay K Baranwal is the Center Director at CDLe (Center for Deep Learning in Electronics Manufacturing). He leads applied data science research and development efforts to solve electronics and semiconductor manufacturing problems. Many of his work at the Center relates to machine vision, learning from limited data, and building digital twins to synthesize new data. Before the Center, he has worked on several TensorFlow-based applications, including a Prediction and Diagnostic system, a Document retrieval, and an information extraction system. He holds multiple patents, is coauthor of industrial papers and has been a speaker at related conferences. He is also a co-author of a book named “What’s new in TensorFlow 2.0.”
Workshop | MLOps & Data Engineering | Intermediate
It’s easy to lose track of which changes gave you the best result when you start exploring multiple model architectures. Tracking the changes in your hyperparameter values, along with code and data changes, will help you build a more efficient model by giving you an exact reproduction of the conditions that made the model better.
In this workshop, you will learn how you can use the open-source tool, DVC, to compare increase reproducibility for two methods of tuning hyperparameters: grid search and random search. We’ll go through a live demo of setting up and running grid search and random search experiments. By the end of the workshop, you’ll know how to add reproducibility to your existing projects.
Tutorial | NLP | Intermediate-Advanced
This tutorial targets AI researchers and practitioners who are interested in applying language processing techniques in cross-domain or cross-lingual tasks. I will provide the audience with a holistic view of (i) a wide selection of representation learning methods for text and multimedia data, (ii) techniques for aligning and transferring knowledge across multiple representations with limited supervision, and (iii) a wide range of applications using these techniques in natural language understanding. Participants will learn about recent trends and emerging challenges in this topic, representative tools and learning resources to obtain ready-to-use models, and how related models and techniques benefit real-world language-related AI applications.
Kai-Wei Chang is an assistant professor in the Department of Computer Science at the University of California Los Angeles (UCLA). His research interests include designing robust machine learning methods for large and complex data and building fair, reliable, and accountable language processing technologies for social good applications. Dr. Chang has published broadly in natural language processing, machine learning, and artificial intelligence. His research has been covered by news media such as Wires, NPR, and MIT Tech Review. His awards include the Sloan Research Fellowship (2021), the EMNLP Best Long Paper Award (2017), the KDD Best Paper Award (2010), and the Okawa Research Grant Award (2018). Dr. Chang obtained his Ph.D. from the University of Illinois at Urbana-Champaign in 2015 and was a post-doctoral researcher at Microsoft Research in 2016. Additional information is available at http://kwchang.net
Half-Day Training | MLOps & Data Engineering | Beginner-Intermediate
Flyte is a Kubernetes native, workflow automation platform for business critical Machine learning and Data science workflows. It enables the user to focus on the business logic, while alleviating the management of infrastructure to a centralized team. It also enables platform teams to provide a self serve platform for their users.
This training should enable you to answer the question if Flyte is right for your organization, what use cases can be best serviced by Flyte.
Ketan Umare is the TSC Chair for Flyte (incubating under LF AI & Data). He is also currently the Chief Software Architect at Union.ai. Previously he had multiple Senior Lead roles at Lyft, Oracle and Amazon ranging from Cloud, Distributed storage, Mapping (map making) and machine learning systems. He is passionate about building software that makes developer and other engineers’ lives easier and provides simplified access to large scale systems. With Flyte he is trying to bridge gap from ideation to productionization for data and ML pipelines and bring a battle tested approach and structure to the data and ML world.
The success of a machine learning project depends on many components: data, algorithms, hardware backend. Developing a machine learning system in production is further complicated by the glaring gap between the development and production environments.
To reduce the overhead cost of communication and collaboration, many companies expect their data scientists and ML engineers to own their projects end-to-end, from data management to modeling to deployment, forcing them to learn tools out of their comfort zones.
In the first part of the workshop, we will cover the challenges of different phases of productionizing machine learning models, as well as the gap between the development and production environments. We will discuss various solutions to address the gap.
In the second half, we will walk over a hands-on tutorial on how to use Metaflow to push the development code from a local machine to production on AWS Batch with a line of code.
Chip Huyen is an engineer and founder working to develop tools that leverage real-time machine learning. Through her work with Snorkel AI, NVIDIA, and Netflix, she has helped some of the world’s largest organizations deploy machine learning systems. She teaches Machine Learning Systems Design at Stanford. She’s also published four bestselling Vietnamese books.
Tutorial | Cybersecurity
Steve is a high-tech investigator and business consultant with over 25 years of experience having worked within a big four accounting firm, a national accounting firm as well as having started his own software and consulting company. He specializes in the utilization of information technology and information analysis within complex corporate disputes, investigations, litigation and business turnarounds. His broad range of experience spans the disciplines in digital forensics, investigations, risk management, cyber security, IT management, data analytics and litigation support.
He has worked on hundreds of engagements, from investigating small IP theft and employee misconduct cases to large complex international Ponzi and fraud schemes where he managed cross boarder teams that collected and analyzed information on matters that often took years to resolve. Steve also serves as a testifying expert on cases
Workshop | Cybersecurity
Workshop | Responsible AI
Neil Sahota is an IBM Master Inventor, United Nations (UN) AI Advisor, author of the book Own the A.I. Revolution., and Chief Innovation Officer at UC Irvine. He is a business solution advisor to several large companies and sought-after keynote speaker. Over his 20+ year career, Neil has worked with enterprises on the business strategy to create next generation products/solutions powered by emerging technology as well as helping organizations create the culture, community, and ecosystem needed to achieve success such as the U.N.’s AI for Good initiative. Neil also actively pursues social good and volunteers with nonprofits. He is currently helping the Zero Abuse Project prevent child sexual abuse as well as Planet Home to engage youth culture in sustainability initiatives.
Lara is a Data Science Manager at EY and occasional adjunct at the University of Chicago’s Booth School of Business, teaching Python and R. Previously she’s taught a data science bootcamp and built risk models for large financial institutions at McKinsey & Co.
Half-Day Training | Machine Learning
Lak is the Director for Data Analytics and AI Solutions on Google Cloud. His team builds software solutions for business problems using Google Cloud’s data analytics and machine learning products. He founded Google’s Advanced Solutions Lab ML Immersion program and is the author of three O’Reilly books and several Coursera courses. Before Google, Lak was a Director of Data Science at Climate Corporation and a Research Scientist at NOAA.
Follow him on Twitter at @lak_gcp, read articles by him on Medium, and see more details at www.vlakshman.com
Noemi Derzsy is a Senior Inventive Scientist at AT&T Chief Data Office within the Data Science and AI Research organization. Her research is centered on understanding and modeling customer behavior and experience through large-scale consumer and network data, using machine learning, network analysis/modeling, spatio-temporal mining, text mining and natural language processing techniques. She is an organizer of Data Umbrella meetup group and NYC Women in Machine Learning and Data Science meetup group, and she is a NASA Datanaut.
Prior to joining AT&T, Noemi was a Data Science Fellow at Insight Data Science NYC and a postdoctoral research associate at Social Cognitive Networks Academic Research Center at Rensselaer Polytechnic Institute. She holds a PhD in Physics, MS in Computational Physics, and has a research background in Network Science and Computer Science.
Workshop | Deep Learning
Jennifer Davis, Ph.D. is a Staff Field Data Scientist at Domino Data Labs, where she empowers clients on complex data science projects. She has completed two postdocs in computational and systems biology, trained at a supercomputing center at the University of Texas, Austin, and worked on hundreds of consulting projects with companies ranging from start-ups to the Fortune 100. Jennifer has previously presented topics for Association for Computing Machinery on LSTMs and Natural Language Generation and at conferences across the US and in Italy. Jennifer was part of a panel discussion for an IEEE conference on artificial intelligence in biology and medicine. She has practical experience teaching both corporate classes and at the college level.
Andrea Lowe, PhD is the Training and Enablement Engineer at Domino Data Labs where she develops training on topics including overviews of coding in Python, machine learning, Kubernetes, and AWS. She trained over 1000 data scientists and analysts in the last year. She has previously taught courses including Numerical Methods and Data Analytics & Visualization at the University of South Florida and UC Berkeley Extension. Her conference experience includes a deep learning tutorial at PyCon, 2 invited talks, 21 poster presentations, and 4 chair positions.
Christopher Crowley has 20 years of experience managing and securing networks, beginning with his first job as an Ultrix and VMS systems administrator at 15 years old. Today, Crowley is a Senior Instructor at the SANS Institute and the course author for SOC-Class.com: the culmination of his thoughts on effective cybersecurity operations. He works with a variety of organizations across industries providing cybersecurity technical analysis, developing and publishing research, sharing expert security insights at conferences, and chairing security operations events. He has provided training to thousands of students globally.
Crowley holds a multitude of cybersecurity industry certifications and provides independent consulting services specializing in effective computer network defense via Montance®, LLC, based in Washington, DC.
Half-Day Training | Cybersecurity
Hannes Hapke works in machine learning at Digits. Prior, he was a senior machine learning scientist for Concur Labs at SAP Concurfor Concur Labs at SAP Concur, where he explored innovative ways to use machine learning to improve the experience of a business traveler. Hannes has also solved machine learning and ML infrastructure problems in various industries including healthcare, retail, recruiting, and renewable energies. He was recognized as a Google Developer Expert for ML and has co-authored two machine learning publications: “Building Machine Learning Pipeline” by O’Reilly Media and “NLP in Action” by Manning Publications.
Workshop | Machine Learning | Intermediate
In this tutorial, I will provide an introduction to the concept of Uplift, compare with traditional response modeling, and review various approaches to Uplift Modeling. The discussion will include approaches to handling a more general situation where only non-experimental (or observational) data are available, integrating causal inference techniques with uplift modeling. We will then discuss the multiple treatment situation to determine the optimal treatment for each individual, which requires constrained optimization using estimates from uplift modeling as inputs. Due to the high uncertainty of lift estimates, various optimization methods will be introduced to handle the uncertainty. While this tutorial is geared towards marketing type applications (“personalized marketing”), the same set of methodologies can be readily applied in other fields such as medicine, insurance, education, political, and social programs. Examples from multiple industries will be used to illustrate its application and methodologies.
Victor has managed teams of quantitative analysts in multiple organizations. He is currently Senior Vice President, Data Science and Artificial Intelligence in Workplace Investing at Fidelity Investments. Previously he managed advanced analytics / data science teams in Personal Investing, Corporate Treasury, Managerial Finance, and Healthcare and Total Well-being at Fidelity Investments. Prior to Fidelity, he was VP and Manager of Modeling and Analysis at FleetBoston Financial (now Bank of America), and Senior Associate at Mercer Management Consulting (now Oliver Wyman).
For academic services, Victor is an elected board member of the National Institute of Statistical Sciences (NISS), where he provides guidance to the board and general education to the statistics community. He has also been a visiting research fellow and corporate executive-in-residence at Bentley University, as well as serving on the steering committee of the Boston Chapter of the Institute for Operations Research and the Management Sciences (INFORMS). Victor earned a master’s degree in Operational Research at Lancaster University, UK, and a PhD in Statistics at the University of Hong Kong, and was a Postdoctoral Fellow in Management Science at University of British Columbia. He has co-authored a graduate level econometrics book and published numerous articles in Data Science, Marketing, Statistics, and Management Science literature. and is co-authoring a graduate-level data science textbook titled “Cause-and-Effect Business Analytics.
Workshop | MLOps & Data Engineering | Machine Learning | Beginner-Intermediate
Machine Learning Operations (MLOps) are essential to build successful Data Science use-cases. Today, ML is powering data driven use-cases that are transforming industries around the world. In order to seize and hold it’s competitive advantage business needs to reduce risk therefore a new expertise rises to include data science models in operational systems.
According to Gartner Research “While many organizations have experimented with AI proofs of concept, there are still major blockers to operationalizing its development. IT leaders must strive to move beyond the POC to ensure that more projects get to production and that they do so at scale to deliver business value. (July 2020)” In this session we will discuss the role of MLOps and how they can help data science models from deployment to maintenance with focus on: keep track of performance degradation overtime from model predictions quality, setting up continuous evaluation metrics and tuning the model performance in both training and serving pipelines that are deployed in production.
Filipa Peleja is the Levi Strauss & Co Europe Lead Data Scientist at the Data Analytics & AI team. She has always been enthusiastic about technology where she first stepped into the tech world as an undergrad in Computer Science and later Ph.D. in the Machine Learning domain. Her academic accomplishments were recognized with the 1st prize of an industry challenge from a telco and publications in international conferences among which, top tier conferences like SIGIR and ACL. Before joining Levi, Filipa interned at Yahoo! Research and, later, worked as a Sr Data Scientist at Vodafone. Filipa loves to work in an area that she feels very passionate about and also enjoys passing along knowledge, hence, she lectures, supervises projects/thesis for CodeOp, Neueda and Barcelona Technical School.
Tutorial | MLOps & Data Engineering | Intermediate
Most of the organizations today use multiple technologies and platforms to develop (and deploy) AI/ML models based on choices and skills of data scientists in different groups. However, while operationalizing those models it is important to still maintain common standards for legal compliance, audit requirements and assurance of overall business value. These requirements run across the areas of Independent Validation of Models; Continuous Monitoring of Model Performance (Accuracy, Bias, Drift, etc) in Production; Model Interpretability; Controlled sourcing and reuse of Data, Features and Packages; Model Facts; Approval steps in Model Lifecycle; Change Management, etc. In this session we shall see how models developed and deployed in Heterogeneous platforms can be still Operationalized keeping above requirements fulfilled but customized based on an organization’s governance requirements. We shall see how Open Source and IBM technologies, can be used to achieve the same with necessary customization. In this session attendees would also learn the technology agnostic key principles those can be used to handle the above mentioned Governance requirements for Operationalization of AI/ML models developed/deployed in heterogeneous platforms.
Sourav Mazumder is a Data Scientist Thought Leader, Open Group Distinguished Data Scientist, responsible for technical thought leadership & strategy, technical vitality, and technical enablement in AI and Data Science area for IBM’s clients and also internal people in IBM. Sourav works with enterprise clients from AI strategy development to implementations and productionization particularly focusing on First of a Kind Projects.
Sourav has consistently driven business innovation and values through Methodologies and Technologies related to Artificial Intelligence, Data Science and Big Data transpired through his knowledge, insights, experience and influencing skills across multiple industries including Manufacturing, Insurance, Telecom, Banking, Media, Health Care and Retail industries in USA, Europe, Australia, Japan and India. In last 10 years he has influenced key decision makers of several fortune 500 companies to adopt Artificial Intelligence, Data Science, and Big Data related technologies to address complex business needs. Sourav has also consistently provided directions to and successfully led numerous challenging Artificial Intelligence, Data Science and Big Data projects, applying various related methodologies ranging from Descriptive statistics, Probabilistic Modeling, Algorithmic Modeling, Natural Language Processing, etc., to solve critical business problems. Sourav has also successfully partnered with academia within North America, India, South Africa to mentor the students and enabling them in this area.
Sourav has experience and exposure in working in variety of Artificial Intelligence, Data Science and Big Data related technologies like Watson Open Scale, Watson Studio, Watson Natural Language Processing, Watson Machine Learning, IBM Cloud Pak for Data, Spark, Hadoop, BigSQL, HBase, MongoDb, Solr, System ML, Brunel, Cognos, R, Python, Scala/Java, etc., using them in projects involving phases from creation of Minimum Viable Product to Productionization. Sourav is an Open Source enthusiast and contributes to Open Source regularly.
Sourav consistently publishes papers/blogs/articles in various industry forums. Sourav is co-author, guest editor and chief editor of multiple books in AI, Data Science and Big Data space. Sourav is regularly invited to speak in various Industry conferences, like Spark Summit, IBM Think, Global AI Conference, etc in this subject area.
Half-Day Training | Deep Learning | Machine Learning | All Levels
PPO is the most common reinforcement learning algorithm when sampling from a simulation can be done quickly and inexpensively. Its magic lies in how it translates hard contraints to losses which enables PPO to simply use standard TensorFlow components. TF-Agents does exactly that, allowing us to play with the most interesting parameters without completely implementing PPO from scratch (even though you will have an idea how to do that after this workshop).
In this slideless workshop we will work our way through a Colab notebook. Along the way you will understand the basic ideas of the PPO reinforcement learning algorithm and how to apply it to a route planning problem. You should be familiar with the basics of machine learning, notebooks, and ideally have worked with TensorFlow 2 before. To let the Colab notebooks run you also need a laptop with a current Chrome browser and a Google account. If you have attended “”Reinforcement Learning with TF-Agents & TensorFlow 2.0: Hands On”” in one of the previous ODSC conferences this would be a natural sequel to it.
Oliver is a software developer from Hamburg Germany and has been a practitioner for more than 3 decades. He specializes in frontend development and machine learning. He is the author of many video courses and textbooks.
Tutorial | Machine Learning | Intermediate
Social networks play a powerful role in shaping our lives, affecting our power and influence, how we acquire and think about basic facts, as well as our behaviors. These effects exist in our professional environments as much as they do in our personal lives. Given the importance of these social structures on our personal and professional lives, it’s incredibly valuable to be able to visualize and analyze them because this knowledge enables you to drive positive change.
Clinton Brownley, Ph.D., is a data scientist at Facebook, where he’s responsible for a variety of analytics projects designed to empower employees to do their best work.
Prior to this role, he was a data scientist at WhatsApp, working to improve messaging and VoIP calling performance and reliability. Before WhatsApp, he worked on large-scale infrastructure analytics projects to inform hardware acquisition, maintenance, and data center operations decisions at Facebook.
As an avid student and teacher of modern analytics techniques, Clinton is the author of two books, “Foundations for Analytics with Python” and “Multi-objective Decision Analysis,” and also teaches Python programming and interactive data visualization courses at Facebook and in the Bay Area.
Clinton is a past-president of the San Francisco Bay Area Chapter of the American Statistical Association and is a council member for the Section on Practice of the Institute for Operations Research and the Management Sciences. Clinton received degrees from Carnegie Mellon University and American University.
Tutorial | Data Analytics | MLOps | All Levels
To really do low latency analytics right we need to take a full-stack approach to the problem. To get the best user experience, the front end, data processing and storage all need to work together but the same time should still allow for a flexible system.
In this workshop we will take a look at each of the parts of a real-time analytics pipeline to understand the options available and trade-offs associated with different technologies and techniques in a modern data pipeline.
Byron has developed large scale data pipelines and processing systems across a variety of industries including Life Sciences, Advertising and Enterprise Software systems. In particular he focuses on distributed systems with low latency requirements for both read and write workloads. Trained as a Statistician with a focus on statistical computing he is also the author of Real-time Analytics published by John Wiley and Sons, which describes both the operational and computational aspects of delivering these systems at scale.
Workshop | Deep Learning | Intermediate
Deepfake synthetic videos and images have found a range of uses, including both harmless and harmful. Fully automated detection is a difficult but important task, particularly since deepfake generation methods continue to evolve. To help people develop successful detection methods, Facebook and other organizations have released public databases of deepfakes to use as training data. This workshop will cover the basic methodologies used for creating deepfakes, the public databases available for training, and some methods (implemented in Python with Keras) for building a detection algorithm. While there are some fully trained detection systems available to web-based users, it is important to know how these systems work and to be able to adjust, adapt, and fine-tune them—skills this workshop aims to equip you with.
Noah Giansiracusa received a PhD in mathematics from Brown University and is an Assistant Professor of Mathematics and Data Science at Bentley University, a business school near Boston. He previously taught at U.C. Berkeley and Swarthmore College. He’s received multiple national grants to fund his research and has been quoted in Forbes, Financial Times, and U.S. News. He is the author of “How Algorithms Create and Prevent Fake News: Exploring the Impacts of Social Media, Deepfakes, GPT-3, and More,” about which Nobel Laureate and former Chief Economist at the World Bank Paul Romer said “There is no better guide to the strategies and stakes of this battle for the future.
Workshop | NLP | Intermediate
In this workshop I will introduce the basics of Natural Language Processing, including the structure of a typical NLP project, with a focus on topic modeling. We will build a topic modeling system using the BBC news dataset. After the workshop you will have a good grasp on the structure of an NLP project, methods used in NLP, and will have built a topic model project by preprocessing and vectorizing the data, building the topic model, visualizing and evaluating it.
Zhenya Antić is an NLP consultant and founder of Practical Linguistics Inc. Her projects include document summarization, information extraction, topic modeling and sentiment analysis of consumer reviews, and document similarity. She is the author of the recently published Python Natural Language Processing Cookbook. Zhenya holds a PhD in Linguistics from the University of California Berkeley and a BS in Computer Science from the Massachusetts Institute of Technology.
Half-Day Training | Deep Learning
Sujit Pal builds intelligent systems around research content that help researchers and medical professionals achieve better outcomes. His areas of interest are Information Retrieval, Natural Language Processing and Machine Learning (including Deep Learning). As an individual contributor in the Elsevier Labs team, he works with diverse product teams to help them solve tough problems in these areas, as well as build proofs of concept at the cutting edge of applied research.