Hong Kong Machine Learning Season 6 Episode 4
23.05.2024 - Hong Kong Machine Learning - ~3 Minutes
When?
- Thursday, May 23, 2024 from 6:15 PM to 9:00 PM (Hong Kong Time)
Where?
- This HKML Meetup is hosted at NCS office, Quarry Bay, Hong Kong.
Thanks to Vahid Asghari (HKML) and Alexandre Gerbeaux (DataRobot) for helping making this event a success!
The page of the event on Meetup: HKML S6E4
Programme:
Talk 1: Business value from Data and Analytics
Speaker: Robin Gallouet is the director leading the Data & Analytics function for Bupa Insurance and Quality HealthCare centres in Hong Kong, overseeing the strategy and execution of data science applications, data platform cloud & engineering, data insights and reporting, data governance and literacy, data automation and the delivery of transformative data projects to permeates through internal operations and customer experience. Robin is an enthusiast, passionate, hands-on leader that specialised the last 5 years on medical insurance, healthcare, and wellness at Bupa in-house, after having spent a decade in EY advisory delivering data science projects for the financial services industry across Paris and Hong Kong. Robin’s laser focus is to bring together an engaged and high-performing team under a meticulously measurable framework to prioritise and assess the value and cost of their time transparently. Robin graduated from french engineering “Grande école” (Groupe Centrale) and further completed his academic interests with econometrics and actuarial sciences.
Talk 2: Building Future-Proof AI-First Analytical Systems: Principles and Insights
Abstract: Analytics answers the question of “What”. Selecting an answer from a set of options is hard, but generating an answer from an unknown set is even harder. Consequently, Large Language Models (LLMs), which are generative in nature, have fundamentally disrupted both software and analytical stacks, necessitating a re-evaluation of how analytical systems are constructed from the ground up. To help people build AI-first analytical system that is future-proof, this talk proposes the main principles of an AI-first Analytical system drawing from theory and the practical insights gained through hands-on experience in AI-generated Analytics.
Speaker: Eason heads an Analytics team on Risk Monitoring Solutions (NLP, Search, Serverless, Microservices) at LRQA; his team mainly serves large global conglomerates, including those from Fortune 500, who face constant ESG risk exposure to their supply chain. Before the role and after business school, Eason mainly served in the industry and worked with startups from the San Francisco and China Bay Area on domains such as AI, Analytics, and Education.
Talk 3: Supervised Cross-Momentum Contrast: Aligning Representations with Prototypical Examples to Enhance Financial Sentiment Analysis
Abstract: In this work, Bo introduces the Supervised Cross-Momentum Contrast (SuCroMoCo) framework, which aims to align financial text representations with prototypical representations based on sentiment categories. This alignment significantly enhances classification performance, particularly in scenarios where pre-trained language models exhibit a limited understanding of the content of financial texts.
Speaker: Bo Peng holds a Ph.D. in Information Technology and Engineering from Yunnan University. Currently, serves as a postdoctoral fellow at the Department of Chinese Bilingual Studies at Hong Kong Polytechnic University. His research interests in Natural Language Processing revolve around representation learning, sentiment analysis, and text generation.