Conf42 Machine Learning 2022

Conf42 Machine Learning 2022 kicks off right now! 🐕 Schedule, Lineup & RSVP: https://www.conf42.com/ml2022 🐈 Join Discord to interact: https://discord.gg/DnyHgrC7jC 0:00 intro, sponsors & partners Keynote 0:40 Jesus Saldana Gonzalez getting started 1:06 Felice Pescatore 2:15 Johannes Hotter 2:43 Vasco Veloso 3:39 Joshua Arvin Lat tools 4:12 Laura Ham 5:04 Julien Simon 5:35 Mohsin Khan 6:20 Andrew Knight lessons learned 6:48 Harika Chebrolu no intro - Aditi Ramaswamy & Anisha Biswaray 7:28 Wojtek Kuberski 8:12 Karan Singh 8:46 Thank you, Join our Discord to interact! https://discord.gg/DnyHgrC7jC

No one can deny that artificial intelligence is all the rage. After a cold winter, it is making a strong comeback with increasingly impressive advances. Maybe that's why people are starting to look with suspicion at this technology. But in this uncertain landscape, one thing is clear: machine learning won't put you out of a job, although you may never work the same way again. Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

Business Agility represents the ability of an organization to respond adequately to market, and at the same time dealing with internal flexibility for revising its organizational model, its own processes and specific skills, in order to make everything more efficient. Business Agility is a process of perpetual evolution that must focus on what is validated on the field in order to identify, from an experimental point of view, the best relative solution, that topically is in conflict with absolutisms. It is evident how the data collected within daily operations represent a real treasure, while emphasizing the importance of turning them into a valuable information that allows decisions to be made in a more targeted and prudent way. To achieve this, modern intelligence algorithms are increasingly used and become a precious ally to those who set themselves the ambitious challenge of implement an agile organization able to best support the vision of business agility. In the time available we will explore just how artificial intelligence can concretely support the organizational transformation process, also presenting the pilot project Arinn.ia, which is a Digital Agile Master that supports teams in their first experiments in agile scope. Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

In this talk, we will focus on the data perspective when building machine learning pipelines. Using two examples, I will show how greenfield and brownfield data labeling differ, what you should focus on in each, and how to best leverage new technologies, frameworks, and products to build high-performing models. The goal is to give you a better understanding of what data options you have for building machine learning pipelines (whether for classification or extraction). The ideas and concepts are based on research results from the Hasso Plattner Institute and three years of experience in consulting AI projects. Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

Books, movies, scientists, researchers... We will look at the history of AI with compelling examples and stories taken from some of the most important periods of this journey towards creating intelligent entities. It will be a great ride! At the end of this talk, you will have a good understanding of the history and reasons behind the research that led to today's features and tomorrow's advancements. Everyone who is curious about Artificial Intelligence should attend. It does not matter if they are practitioners or just want to know a bit more. Understanding the past is important to better shape the future. Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

Designing and building machine learning systems require a lot of skill, time, and experience. Data scientists, developers, and ML engineers work together in building ML systems and pipelines that automate different stages of the machine learning process. Once the ML systems have been set up, these systems need to be secured properly to prevent these systems from being hacked and compromised. Some attacks have been customized to take advantage of vulnerabilities present in certain libraries. Other attacks may take advantage of vulnerabilities present in the custom code of ML engineers as well. There are different ways to attack machine learning systems and most data science teams are not equipped with the skills required to secure the systems they built. In this talk, we will discuss in detail the **cybersecurity attack chain** and how this affects a company's strategy when setting up different layers of security. We will discuss the different ways ML systems can be attacked and compromised and along the way, we will share the relevant strategies to mitigate these attacks. Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

In machine learning, e.g. recommendation tools or data classification, data is often represented as high-dimensional vectors. These vectors are stored in so-called vector databases. With vector databases you can efficiently run searching, ranking and recommendation algorithms. Therefore, vector databases became the backbone of ML deployments in industry. This session is all about vector databases. If you are a data scientist or data/software engineer this session would be interesting for you. You will learn how you can easily run your favorite ML models with the vector database Weaviate. You'll get an overview of what a vector database like Weaviate can offer: such as semantic search, question answering, data classification, named entity recognition, multimodal search, and much more. After this session, you are able to load in your own data and query it with your preferred ML model! Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

As amazing as state-of-the-art machine learning models are, training, optimizing, and deploying them remains a challenging endeavor that requires a significant amount of time, resources, and skills, all the more when different languages are involved. Unfortunately, this complexity prevents most organizations from using these models effectively, if at all. Instead, wouldn’t it be great if we could just start from pre-trained versions and put them to work immediately? This is the exact challenge that Hugging Face is tackling. Founded in 2016, this startup based in New York and Paris makes it easy to add state-of-the-art Transformer models to your applications. Thanks to popular open-source libraries (transformers, tokenizers, and datasets libraries, developers can easily work with over 2,900 datasets and over 29,000 pre-trained models in 160+ languages. In fact, with close to 60,000 stars on GitHub and 1 million downloads per month, the transformers library has become the de-facto place for developers and data scientists to find state-of-the-art models for natural language processing, computer vision, and audio. In this session, we'll introduce you to Transformer models and what business problems you can solve with them. Then, we'll show you how you can simplify and accelerate your machine learning projects end-to-end: experimenting, training, optimizing, and deploying. Along the way, we'll run some demos to keep things concrete and exciting! Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

Defect and anomaly detection during manufacturing processes is a vital step to ensure the quality of the products. Amazon Lookout for Vision is a machine learning (ML) service that spots defects and anomalies in visual representations using computer vision (CV). With Amazon Lookout for Vision, manufacturing companies can increase quality and reduce operational costs by quickly identifying differences in images of objects at scale. This session will walk through Amazon Lookout for Vision, and how it can be integrated with other AWS Serverless services to automate defect detection, get real-time alerts and visualize business insights from the process. Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

Apps break. It's what they do. Everyone has seen skewed layouts, missing buttons, and overlapping text. Those visual bugs are such a pain because traditional test automation usually can't catch them. However, they cause serious reputational risk because appearances matter. The types of problems scripts can catch usually require complicated element locators and assertions, too. The best way to catch visual problems is to look at them with human eyes. People are good at quickly noticing things that don't look right. If we can train an AI model to look for important visual differences between app snapshots, then we can automate visual testing! In this talk, I'll show how how to apply AI-backed visual comparisons to end-to-end test automation. We'll transform traditional tests into much simpler scenarios that save time for both development and execution. You'll see how to make visual comparisons between baselines and updated snapshots. A picture is truly worth a thousand assertions. Ultimately, visual testing like this enables you to spend more time on proper test coverage and less time on automation implementation! Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

The major challenge is that investors/policy makers/railways/electrical department don't have the enough information to come up with viable solar energy investments which are sustainable and cost-effective hence, high potential clean energy projects remain unidentified and thus not getting deployed. By making visible the energy and costs savings potential, more projects will be deployed and contribute to a greener energy system thereby we can reduce global warming. One such method is to install solar panels and harness the energy from the sun. We will detect rooftops and give understandable rooftops classification thus, accelerate the growth of solar installations in a given area in order to identify the potential of facilities’ solar installation depending on the uncluttered surface area, shading, direction, material and location. The above data points can be used as input into a detailed building energy simulation. The goal of the project is to develop a production-ready deep vision engine to provide accurate rooftop solar PV analysis so that the platform operates across building portfolios and thereby helps the property owners to identify and prioritize top candidates for solar PV and battery installations in terms of return on investment and carbon emission reduction. Furthermore, today’s techniques are susceptible to noise from varying bottom conditions and climatic conditions, shadowing. And the current ecosystem is providing data about the amount of energy production based on solar panels installed on rooftops. Here, in the present idea, we will spot optimal locations for solar panels installation on the rooftop depending on shadowing and direction, amount of usage, surroundings, etc. Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

Amassing large amounts of data is in the very nature of most tech companies. However, in companies which don't have a dedicated ML focus, there is often relatively little effort put into thoroughly analysing this data and finding uses for it. In fact, there exists a misconception that ML is ""irrelevant"" to certain problems, and thus spending time on ML data analysis is a waste. The real waste, as Anisha and I discovered, lies in ignoring this data. Through using our large body of historical data and implementing a simple random forest regression, we were able to discover a method of routing Lob mailpieces which, as time will tell, could have a significant impact on our mail delivery timings. Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

The objective of the talk is to build understanding on why and how you need to monitor ML in production. We'll cover the taxonomy of failures based on use cases, data, characteristics of systems they interact with, and human involvement. You'll learn the tools (both statistical and algorithmic) used in dealing with these failures, their applications, and their limits. Fact is, the world changes and data drift and concept drift lead to model degradation and losses to the business. We'll leverage real-life use cases to showcase the importance of ML monitoring in one of the biggest industries. Finally, we'll show you how to address this by by monitoring ML performance Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk

Smart Cities are complex and challenging environments. They generate an overwhelming amount of data that you have to ingest, transfer, prepare and store even before thinking of analyzing them or training a model. In this demo, we will explain how we architected and deployed a variety of data engineering patterns to create a full edge-to-core data & machine learning pipeline on OpenShift/Kubernetes for a smart-city use case. We will walk you through the approach we took to deploy the ML model on K8s, moving data from edge-to-core using kafka, creation of data aggregation pipelines, and demo of real-time and batch analysis etc. Our overarching goal was to possess the ability to re-deploy the entire stack with a single command, for which we used Ansible and we lived happily thereafter. By the end of this session, you should get a better understanding on how to architect and develop data engineering workflows, and how to automate the deployment of the entire stack using Ansible. Other talks at this conference 🚀🪐 https://www.conf42.com/ml2022 — 0:00 Intro 0:22 Talk