Schedule: Models and Methods sessions: Artificial Intelligence Conference: Applied AI & machine learning

11:05–11:45 Wednesday, 16 October 2019

Location: Buckingham Room - Palace Suite

Introducing a new anomaly-detection algorithm (SR-CNN) inspired by computer vision

Secondary topics: Computer Vision, Data, Data Networks, Data Quality, Deep Learning, Machine Learning, Temporal data and time-series

Qun Ying (Microsoft)

Average rating:

(5.00, 2 ratings)

Anomaly detection may sound old fashioned, yet it's super important in many industry applications. Tony Xing, Bixiong Xu, Congrui Huang, and Qun Ying detail a novel anomaly-detection algorithm based on spectral residual (SR) and convolutional neural network (CNN) and explain how this method was applied in the monitoring system supporting Microsoft AIOps and business incident prevention. Read more.

11:55–12:35 Wednesday, 16 October 2019

Location: King's Suite - Sandringham

Sequence to sequence (S2S) modeling for time series forecasting

Secondary topics: Deep Learning, Machine Learning, Temporal data and time-series

Arun Kejariwal (Independent), Ira Cohen (Anodot)

Average rating:

(4.00, 5 ratings)

Sequence to sequence (S2S) modeling using neural networks has become increasingly mainstream in recent years. In particular, it's been used for applications such as speech recognition, language translation, and question answering. Arun Kejariwal and Ira Cohen walk you through how S2S modeling can be leveraged for these use cases, visualization, real-time anomaly detection, and forecasting. Read more.

11:55–12:35 Wednesday, 16 October 2019

Location: Buckingham Room - Palace Suite

Deep learning with TensorFlow Probability in cancer prediction with reporting confidence

Secondary topics: Computer Vision, Deep Learning, Health and Medicine, Machine Learning

Biraja Ghoshal (Tata Consultancy Service)

Average rating:

(1.00, 2 ratings)

Deep learning, which involves powerful black box predictors, has achieved state-of-the-art performance in medical imaging analysis, such as segmentation and classification for diagnosis, but knowing how much confidence there is in a prediction is essential for gaining clinicians' trust. Biraja Ghoshal explores probabilistic modeling with TensorFlow Probability in cancer prediction. Read more.

11:55–12:35 Wednesday, 16 October 2019

Location: Blenheim Room - Palace Suite

Building differentially private machine learning models using TensorFlow

Secondary topics: Ethics, Security, and Privacy, Machine Learning, Machine Learning tools

Chang Liu (Georgian Partners ), Ji Chao Zhang (Georgian Partners)

Average rating:

(4.33, 3 ratings)

The world is increasingly data driven, and people have developed an awareness and concern for their data. Chang Liu and Ji Chao Zhang examine differential privacy—the component of the TensorFlow Privacy library that allows users to train differentially private logistic regression and support vector machines—along with real-world use cases and demonstrations for how to apply the tools. Read more.

14:35–15:15 Wednesday, 16 October 2019

Location: Blenheim Room - Palace Suite

Extracting trading signals from alternative data using machine learning

Secondary topics: Machine Learning, Text, Language, and Speech

Arun Verma (Bloomberg)

Average rating:

(3.50, 4 ratings)

To gain an edge in the markets, quantitative hedge fund managers require automated processing to quickly extract actionable information from unstructured and increasingly nontraditional sources of data. Arun Verma shares NLP, AI, and ML techniques that help extract derived signals that have significant trading alpha or risk premium and lead to profitable trading strategies. Read more.

16:00–16:40 Wednesday, 16 October 2019

Location: Buckingham Room - Palace Suite

Audience projection of target consumers over multiple domains: A NER and Bayesian approach

Secondary topics: Data, Data Networks, Data Quality, Machine Learning, Text, Language, and Speech

Gianmario Spacagna (Helixa)

Average rating:

(4.00, 2 ratings)

AI-powered market research is performed by indirect approaches based on sparse and implicit consumer feedback (e.g., social network interactions, web browsing, or online purchases). These approaches are more scalable, authentic, and suitable for real-time consumer insights. Gianmario Spacagna proposes a novel algorithm of audience projection able to provide consumer insights over multiple domains. Read more.

16:00–16:40 Wednesday, 16 October 2019

Location: Blenheim Room - Palace Suite

Adversarial network for natural language synthesis

Secondary topics: Deep Learning, Ethics, Security, and Privacy, Machine Learning, Reinforcement Learning, Text, Language, and Speech

Rajib Biswas (Ericsson)

Average rating:

(4.00, 2 ratings)

Rajib Biswas outlines the application of AI algorithms like generative adversarial networks (GANs) to solve natural language synthesis tasks. Join in to learn how AI can accomplish complex tasks like machine translation, write poetry with style, read a novel, and answer your questions. Read more.

9:30–9:45 Thursday, 17 October 2019

Location: King's Suite

The quest for high-quality data

Secondary topics: Data, Data Networks, Data Quality, Machine Learning

Ihab Ilyas (University of Waterloo)

Average rating:

(4.30, 10 ratings)

Ihab Ilyas highlights the data-quality problem and describes the HoloClean framework, a state-of-the-art prediction engine for structured data with direct applications in detecting and repairing data errors, as well as imputing missing labels and values. Read more.

11:05–11:45 Thursday, 17 October 2019

Location: King's Suite - Balmoral

Principled tools for analyzing weight matrices of production-scale deep neural networks

Secondary topics: Deep Learning, Deep Learning tools

Michael Mahoney (UC Berkeley)

Average rating:

(3.00, 4 ratings)

Developing theoretically principled tools to guide the use of production-scale neural networks is an important practical challenge. Michael Mahoney explores recent work from scientific computing and statistical mechanics to develop such tools, covering basic ideas and their use for analyzing production-scale neural networks in computer vision, natural language processing, and related tasks. Read more.

11:05–11:45 Thursday, 17 October 2019

Location: King's Suite - Sandringham

Online evaluation of machine learning models

Secondary topics: Machine Learning

Ted Dunning (MapR, now part of HPE)

Average rating:

(4.50, 4 ratings)

Evaluating machine learning models is surprisingly hard, but it gets even harder because these systems interact in very subtle ways. Ted Dunning breaks the problem into operational and functional concerns and shows you how each can be done without unnecessary pain and suffering. You'll also get to see some exciting visualization techniques to help make the differences strikingly apparent. Read more.

11:55–12:35 Thursday, 17 October 2019

Location: King's Suite - Sandringham

A pragmatic introduction to building NLP models

Secondary topics: Data, Data Networks, Data Quality, Machine Learning, Machine Learning tools, Text, Language, and Speech

Julien Simon (AWS)

Average rating:

(4.86, 7 ratings)

Many natural language processing (NLP) tasks require each word in the input text to be mapped to a vector of real numbers. Julien Simon explores word vectors, why they’re so important, and which are the most popular algorithms to compute them (Word2Vec, GloVe, BERT). You'll get to see how to solve typical NLP problems through several demos by either computing embeddings or reusing pretrained ones. Read more.

13:45–14:25 Thursday, 17 October 2019

Location: Blenheim Room - Palace Suite

Anomaly detection using deep learning to measure the quality of large datasets

Secondary topics: Data, Data Networks, Data Quality, Deep Learning tools, Machine Learning, Temporal data and time-series

Sridhar Alla (BlueWhale)

Average rating:

(1.67, 3 ratings)

Any business, big or small, depends on analytics, whether the goal is revenue generation, churn reduction, or sales or marketing purposes. No matter the algorithm and the techniques used, the result depends on the accuracy and consistency of the data being processed. Sridhar Alla examines some techniques used to evaluate the quality of data and the means to detect the anomalies in the data. Read more.

14:35–15:15 Thursday, 17 October 2019

Location: King's Suite - Balmoral

Concepts and tools for fairness, explainability, and robustness in machine learning

Secondary topics: Ethics, Security, and Privacy, Machine Learning

Ilya Feige (Faculty)

Average rating:

(5.00, 2 ratings)

Ilya Feige explores AI safety concerns—explainability, fairness, and robustness—relevant for machine learning (ML) models in use today. With concepts and examples, he demonstrates tools developed at Faculty to ensure black box algorithms make interpretable decisions, do not discriminate unfairly, and are robust to perturbed data. Read more.

14:35–15:15 Thursday, 17 October 2019

Location: Blenheim Room - Palace Suite

A practical guide toward algorithmic bias and explainability in machine learning

Secondary topics: Ethics, Security, and Privacy, Machine Learning

Alejandro Saucedo (The Institute for Ethical AI & Machine Learning)

Average rating:

(4.00, 4 ratings)

Alejandro Saucedo demystifies AI explainability through a hands-on case study, where the objective is to automate a loan-approval process by building and evaluating a deep learning model. He introduces motivations through the practical risks that arise with undesired bias and black box models and shows you how to tackle these challenges using tools from the latest research and domain knowledge. Read more.

16:00–16:40 Thursday, 17 October 2019

Location: Buckingham Room - Palace Suite

NLP for healthcare: Feature engineering and model diagnostics

Secondary topics: Data, Data Networks, Data Quality, Health and Medicine, Machine Learning, Text, Language, and Speech

Manas Ranjan Kar (Episource)

Natural language processing (NLP) is hard, especially for clinical text. Manas Ranjan Kar explains the multiple challenges of NLP for clinical text and why it's so important that we invest a fair amount of time on domain-specific feature engineering. It’s also crucial to understand to diagnose an NLP model performance and identify possible gaps. Read more.

16:00–16:40 Thursday, 17 October 2019

Location: Blenheim Room - Palace Suite

Anomaly detection in smart buildings using federated learning

Secondary topics: Data, Data Networks, Data Quality, Ethics, Security, and Privacy, Machine Learning, Machine Learning tools, Mobile Computing, IoT, Edge

Tuhin Sharma (Binaize), Bargava Subramanian (Binaize)

Average rating:

(4.50, 2 ratings)

There's an exponential growth in the number of internet-enabled devices on modern smart buildings. IoT sensors measure temperature, lighting, IP camera, and more. Tuhin Sharma and Bargava Subramanian explain how they built anomaly-detection models using federated learning—which is privacy preserving and doesn't require data to be moved to the cloud—for data quality and cybersecurity. Read more.

16:50–17:30 Thursday, 17 October 2019

Location: Buckingham Room - Palace Suite

An artificial intelligence framework to counter international human trafficking

Secondary topics: Ethics, Security, and Privacy, Machine Learning, Text, Language, and Speech

Tom Sabo (SAS)

Average rating:

(4.50, 2 ratings)

Efforts to counter human trafficking internationally must assess data from a variety of sources to determine where best to devote limited resources. Tom Sabo explores text-based machine learning, rule-based text extraction to generate training data for modeling efforts, and interactive visualization to improve international trafficking response. Read more.