Workshop to be announced
Workshop Details
Dates
To be Announced
Duration
To be Announced
Timings
To be Announced
Speaker
To be Announced
Venue
Online
Workshop Syllabus
Workshop Syllabus
Introduction to AI
- Classification and regression trees vs. neural networks (NN): pros and cons
- How to train a NN: backpropagation and stochastic gradient descent
- Types of NN: multilayer perceptron (MLP), recurrent neural network (RNN), convolutional neural network (CNN), and related variants
- Discriminative vs. generative AI: a Bayesian perspective, and what generative models can do that discriminative models can’t
- Exercise: Build, train, and apply an RNN to predict SPX returns with the help of a chatbot such as ChatGPT
Deep Autoregressive Models and Transformers
- Deep Autoregressive Models and TransformersProbabilistic modeling of time series: reducing model complexity and dimensionality
- Limitations of RNNs: slow training, vanishing/exploding gradients, high memory usage
- Transformers to the rescue: Attention Is All You Need
- Exercise: Build, fine-tune, and apply the Lag-Llama transformer to predict exchange rates
LLMs for Sentiment Analysis in Trading
- BERT and FinBERT: fine-tuning a pre-trained LLM with financial text data; what pre-training and fine-tuning mean, and how to fine-tune
- Embeddings: converting English text into numerical input for a NN
- Exercise: Use FinBERT to compute sentiment scores on Fed Chair speech transcripts and backtest a trading model on SPY based on these scores
