A neural network architecture that relies on self-attention mechanisms to process sequential data. Transformers excel at understanding context by relating different parts of the input sequence to each other. Originating in language models (e.g. BERT, GPT), transformers have achieved state-of-the-art results in NLP and are now applied to time series as well. In forex trading, transformer-based models can be used for tasks like news analysis (understanding the context of financial reports) or even for forecasting by capturing long-range dependencies in price data. Their ability to process entire sequences in parallel makes them powerful for extracting features from large volumes of data in modern AI trading systems.

Back to Robots Glossary Index