Skip to content

Details

Step into a thrilling exploration of financial data science and generative AI at our upcoming meetup! Unravel the secrets of Time-based Cross Validation, a potent tool for effective handling of time-sensitive financial data. Decode the US Treasury yield curve through a comprehensive PCA Decomposition, getting an in-depth insight into the dynamics of our economy. Finally, delve into Practical GenAI, understanding the true power of Language Learning Models (LLMs) beyond mere prompt engineering. Discover the art of fine-tuning these models and leveraging techniques like quantization for scaling. Ignite your passion for data and AI with us!

Pizzas and refreshments will be provided (thanks Fullerton Fund Management!)

Agenda

6.30pm - 7.00pm: Registration + Refreshment

7.00pm - 7.10pm: Welcome and Intro by Chao Jen | Zhao Ren

7.10pm - 7.30pm: Time-based Cross Validation for Financial Data by Shi Hui

7.30pm - 7.50pm: PCA Decomposition of the UST Yield Curve by Yan Rong

7.50pm - 8.20pm: Practical GenAI - Fine-tuning to create domain-specific LLMs by Kai Xin

8.20pm - 8.30pm: Q&A and closing

Synopsis

Time-based Cross Validation for Financial Data

When dealing with financial data, the time component is an important aspect that must be respected. In a typical cross-validation, random samples are chosen and assigned to either train or test. However, this cannot be done for data with a time dimension. In this segment, we will explore how you can overcome this issue by introducing a variation of cross validation, known as Time-based CV which can be applied to any time-based data, even those where we have multiple observations per timestamp.

PCA Decomposition of the UST Yield Curve

A yield curve is the cross-sectional relationship between the bond maturities and its yields at a given point in time, and the US Treasury yield curve has been used as a tool for investors to get a sense of where the economy may be headed. In this session, we will look at a PCA decomposition of the UST yield curve using PCA and analyze the sensitivities of 4 different points of the curve (2y, 5y, 10y, 30y) to each PC.

Practical GenAI - Fine-tuning to create domain-specific LLMs

We will explore how to scale GenAI and create powerful AI products efficiently. We will start by delving into Prompt Engineering, generating technical analysis with simple prompt engineering, and leveraging zero/one/few-shot learning with Retrieval Augmented Generation (RAG). Next, we will explore the exciting realm of fine-tuning LLMs, moving beyond the limitations of public API solutions. We'll highlight advanced techniques like Instruction-Tuning, Adapters and PEFT-LoRA with Quantization that can enhance efficiency and control, creating a "Swappable LLM Burger Stack".

Speaker Bio(s):

Shi Hui is a data scientist with a passion for unraveling the mysteries of the stock market through data-driven insights. She has been with Fullerton Fund Management Company for over two years and is responsible for driving Augmented Intelligence initiatives within the Equities team. She focuses on applying Machine Learning and Natural Language Processing models in her role and has recently started to explore large language model applications for investment research.

Yan Rong is a data scientist focusing on Fixed Income in the investment department of Fullerton Fund Management Company. She seeks to provide fresh perspective to the team through her research and looks forward to being able to continue grow and learn in her role, through which she has experienced in full the saying "We learn something new everyday". Embracing the opportunity to step outside her comfort zone, she will make her first presentation as a data scientist in the DSSG meetup.

Kai Xin works at the intersection of data and product innovation. He has over a decade of experience creating machine learning products such as deep learning sentiment models, knowledge graphs, recommender systems, segmentation & targeting across industries like finance, media, jobs, eCommerce, and healthcare. He co-founded DataScience SG.

Related topics

Events in Singapore, SG
Artificial Intelligence
Machine Learning
Data Analytics
Data Engineering
Open Source

You may also like