2024-06-13 –, Room A - IJzaal
At Emakia, we are developing a machine learning system aimed at protecting social media users from harassment. Our project is in the development phase, with a focus on providing immediate support to individuals targeted online. Our model not only screens harassing content but also links victims to supportive resources catered to diverse groups, including LGBTQ+ communities and minority networks.
We hold two patents for this innovative system and use Google’s Vertex AI and Apple’s Core ML to refine our models. Google has supported our efforts with $100,000 worth of services, which has enabled extensive training and data acquisition. We also contribute to the Women Who Code community by sharing updates on our progress.
In collaboration with Montclair NLP Lab, we stress the importance of Trustworthy AI, ensuring that our data inputs and model predictions are both reliable and ethical. Plans are underway to establish a data hub to facilitate academic research and provide resources to the scientific community.
MozFest Session Overview:
Part 1: Emakia Tech System Demonstration
• Introduction to our system’s capabilities, demonstrating our approach to protecting users from online harassment.
• Discussion on our MLOps framework, detailing the processes for deploying and managing ML models.
o Google Cloud Development with Vertex AI Node.js
Label validation.
Model training and deployment.
Creating pipelines with real-time Twitter data.
Model output evaluation.
o Apple Development with Core ML
Emphasis on model training and deployment for iOS applications.
Part 2: Collaborative Workshop with MozFest Participants
• Interactive sessions with Miro tool to engage in discussions on:
1. Model and Data Evaluation
Validation of label data.
Real-time data model assessment.
2. Transparency in AI
Promoting open-source practices.
Developing a transparent Data Hub.
Managing software and data set licensing.
3. Legal Responsibility
Discussing legal strategies when models do not achieve 100% accuracy.
This session will highlight our technological progress and involve participants in crucial dialogues on the future of AI in mitigating online harassment. Our goal is to create a safer digital environment, especially for vulnerable groups such as women, minorities, and the LGBTQ+ community.
Corinne serves as the CEO/CTO of Emakia and Emakia Tech, boasting a multifaceted career as an inventor, engineer, and data scientist with a keen focus on AI innovation. With over twenty-six years of experience in engineering across diverse technologies, she consistently demonstrates expertise and innovation.
Over the last five years, Corinne has spearheaded Emakia Tech's growth, notably contributing to their flagship application, Enaëlle. Leveraging advanced Machine Learning tools like Vertex AI and Core ML, she engineered a system adept at filtering harassment from social media content, showcasing her ability to address pressing societal issues through AI.
Corinne's impact extends beyond Emakia Tech. Previously, she developed AI software predicting solar power production capacity and contributed to protein identification prediction from electrophoresis images, bridging AI with life sciences. At Molecular Devices, she enhanced the Softmaxpro software, showcasing her meticulous attention to detail and problem-solving skills.
Her innovative solutions led to two patent applications for the Emakia system, highlighting her ability to pioneer groundbreaking AI technologies. Armed with a Bachelor of Science in computer science from San Francisco State University, Corinne combines academic prowess with industry expertise.
Her dedication to advancing AI, coupled with her knack for tackling challenges and devising practical solutions, positions her as a thought leader in the field. Corinne's talk promises to offer valuable insights into the societal impact of AI, making her a compelling speaker with a wealth of knowledge and hands-on experience.