Please use this identifier to cite or link to this item:
|Title:||Foreign exchange prediction and trading using few-shot machine learning||Authors:||Lee, Wilson||Keywords:||Engineering::Electrical and electronic engineering||Issue Date:||2021||Publisher:||Nanyang Technological University||Source:||Lee, W. (2021). Foreign exchange prediction and trading using few-shot machine learning. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/150115||Abstract:||Forecasting is one of many machine learning applications for time-series data. However, forecasting market prices in forex and the stock market remain a challenge despite extensive research in several state-of-the-art machine learning methods While majority of the market moves within a range for a given period, the many variables incorporated into the trading market make short-term movements highly unpredictable and difficult to forecast with high accuracy solely from past performance. Hence, this paper seeks to determine the usefulness of Few Shot Learning on time-series data. This paper explores 2 different approaches related to Few Shot; looking into classifying trends and transfer learning on small datasets. In the first approach, we sub-sample prices and convert them into a line graph image database, classifying them into distinct feature classes. This allows the model to learn useful patterns to be applied on a target dataset with limited data for forecasting. The second approach uses transfer learning, by taking well-defined models and training them with large amount of time-series data. The trained model is then used to forecast on a small target dataset. The results are compared against benchmarks from few-shot learning classification techniques and LSTM models that have been published in the past. The result shows possible effectiveness of these approaches that can help improve the accuracy and reduce reliance on having large target dataset.||URI:||https://hdl.handle.net/10356/150115||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Student Reports (FYP/IA/PA/PI)|
Updated on Dec 4, 2022
Updated on Dec 4, 2022
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.