Date: 18th January 2025
Time: 11:00 AM – 01:00 PM IST
Huge shoutout to BrowserStack, our Exclusive sponsor for all community events and Premier sponsor for Conferences. We thank them for supporting this Event.
As machine learning models continue to shape critical decisions in areas like healthcare, finance, and security, understanding their vulnerabilities has become paramount. “Breaking the Machine” delves into adversarial attacks—carefully crafted actions designed to exploit model weaknesses, leading to incorrect predictions. This talk explores the two main categories of adversarial attacks, White Box and Black Box, and their subcategories of targeted and untargeted attacks. We’ll also explore behavioral attacks, including Minimum Functionality Tests (MFTs), Invariance Tests (INV), and Directional Expectation Tests (DETs), which examine a model’s robustness and reliability. Beyond attacks, we’ll explore various biases that can affect a model’s fairness and accuracy during development and deployment.
Key Takeaways:
With real-world examples, this session offers a comprehensive overview of how adversarial attacks exploit model weaknesses and how bias affects the reliability of AI systems. By the end, you’ll gain insight into strengthening models against these threats and ensuring they perform reliably across diverse environments and populations.
Software Engineer
Deepika Hanumanthu is a passionate about software testing with over six years of expertise. She discovered her passion for machine learning during her master’s at the University of Stuttgart, Germany.
Deepika’s groundbreaking research in machine learning has transformed testing methodologies, especially in tackling adversarial attacks and bias.
As a specialist in crafting robust Test Automation frameworks with Python and Pytest, she has elevated quality and efficiency at ifm. Her skills encompass comprehensive UI, API, and Penetration testing, ensuring flawless execution from Functional end-to-end Testing to Regression and Exploratory Testing.
Managing Test Data is one of the most challenging aspects of testing and software development, especially as systems grow in complexity n require intricate, context specific data…..The need to generate, anonymize, and transform data at scale often consumes significant testing time and resources, which in turn impacts efficiency.
However, without relevant test data, we cannot effectively trigger actions or observe system behaviors.
LLM offer a powerful solution to this challenge by generating diverse and complex data structures on demand.
By leveraging carefully designed prompts we can streamline data creation, reduce manual effort, and focus on higher-value testing activities.
Key Takeaways
Simplify Test Data Management enabling faster, scalable, and more accurate test execution.
QA Manager at Clover Bay Technologies
Komal Chowdhary is a Self driven Quality evangelist with a decade of industry experience
The Test Tribe is a leading global Software Testing Community (proudly Asia’s Largest) turned EdTech Startup. Started in 2018 with a mission to give Testing Craft the glory it deserves while we co-create Smarter, Prouder, more confident Testers.
We take pride in creating unique global Events, Online Community spaces, and eLearning platforms where Software Testers across the globe collaborate, learn and grow.
With around 200 Software Testing Events like Conferences, Hackathons, Meetups, Webinars, etc., and with other Community initiatives, we have reached a global footprint of over 70K+ Testers.
We intend to be a one-stop destination of choice for Testers across the globe for their upskilling and community needs.
