Time series problems are currently of high interest and they occur in a wide variety of fields ranging from engineering, social sciences, healthcare, climate analysis and finance among others.
Time Series inference involves analyzing data which evolves over time to identify patterns, make predictions, and draw insights about the underlying generating processes. Machine Learning and Statistical algorithms used to perform these tasks help in anticipating critical events like financial risks, improve patient care with predictive diagnostics, and support environmental conservation efforts through accurate predictions of ecological changes.
Our research in this area involves development of novel methods for forecasting of nonstationary time series including both future values and the uncertainty of prediction. We are also interested in developing techniques for classifying and clustering time series for applications such as motif and anomaly detection. Additionally, we are exploring innovative methods for synthetic data generation in time series, which is critical for tasks where generating real-life data can be both expensive and challenging. Our goal is to make a significant impact across various sectors that rely on time series data by providing robust, scalable solutions for complex inference problems.