skip to main content

Title: Transformer Hawkes Process
Modern data acquisition routinely produce massive amounts of event sequence data in various domains, such as social media, healthcare, and financial markets. These data often ex- hibit complicated short-term and long-term temporal dependencies. However, most of the ex- isting recurrent neural network-based point process models fail to capture such dependencies, and yield unreliable prediction performance. To address this issue, we propose a Transformer Hawkes Process (THP) model, which leverages the self-attention mechanism to capture long- term dependencies and meanwhile enjoys computational efficiency. Numerical experiments on various datasets show that THP outperforms existing models in terms of both likelihood and event prediction accuracy by a notable margin. Moreover, THP is quite general and can incorpo- rate additional structural knowledge. We provide a concrete example, where THP achieves im- proved prediction performance for learning multiple point processes when incorporating their relational information.
Authors:
; ; ; ;
Award ID(s):
1717916
Publication Date:
NSF-PAR ID:
10162616
Journal Name:
International Conference on Machine Learning
Sponsoring Org:
National Science Foundation
More Like this
  1. Estimating the future event sequence conditioned on current observations is a long-standing and challenging task in temporal analysis. On one hand for many real-world problems the underlying dynamics can be very complex and often unknown. This renders the traditional parametric point process models often fail to fit the data for their limited capacity. On the other hand, long-term prediction suffers from the problem of bias exposure where the error accumulates and propagates to future prediction. Our new model builds upon the sequence to sequence (seq2seq) prediction network. Compared with parametric point process models, its modeling capacity is higher and has better flexibility for fitting real-world data. The main novelty of the paper is to mitigate the second challenge by introducing the likelihood-free loss based on Wasserstein distance between point processes, besides negative maximum likelihood loss used in the traditional seq2seq model. Wasserstein distance, unlike KL divergence i.e. MLE loss, is sensitive to the underlying geometry between samples and can robustly enforce close geometry structure between them. This technique is proven able to improve the vanilla seq2seq model by a notable margin on various tasks.
  2. According to the National Academies, a week long forecast of velocity, vertical structure, and duration of the Loop Current (LC) and its eddies at a given location is a critical step toward understanding their effects on the gulf ecosystems as well as toward anticipating and mitigating the outcomes of anthropogenic and natural disasters in the Gulf of Mexico (GoM). However, creating such a forecast has remained a challenging problem since LC behavior is dominated by dynamic processes across multiple time and spatial scales not resolved at once by conventional numerical models. In this paper, building on the foundation of spatiotemporal predictive learning in video prediction, we develop a physics informed deep learning based prediction model called—Physics-informed Tensor-train ConvLSTM (PITT-ConvLSTM)—for forecasting 3D geo-spatiotemporal sequences. Specifically, we propose (1) a novel 4D higher-order recurrent neural network with empirical orthogonal function analysis to capture the hidden uncorrelated patterns of each hierarchy, (2) a convolutional tensor-train decomposition to capture higher-order space-time correlations, and (3) a mechanism that incorporates prior physics from domain experts by informing the learning in latent space. The advantage of our proposed approach is clear: constrained by the law of physics, the prediction model simultaneously learns good representations for frame dependenciesmore »(both short-term and long-term high-level dependency) and inter-hierarchical relations within each time frame. Experiments on geo-spatiotemporal data collected from the GoM demonstrate that the PITT-ConvLSTM model can successfully forecast the volumetric velocity of the LC and its eddies for a period greater than 1 week.« less
  3. Product search serves as an important entry point for online shopping. In contrast to web search, the retrieved results in product search not only need to be relevant but also should satisfy customers' preferences in order to elicit purchases. Starting from the same query, customers may purchase different products due to their personal taste or needs. Previous work has shown the efficacy of purchase history in personalized product search. However, customers with little or no purchase history do not benefit from personalized product search. Furthermore, preferences extracted from a customer's purchase history are usually long-term and may not always align with her short-term interests. Hence, in this paper, we leverage clicks within a query session, as implicit feedback, to represent users' hidden intents, which further act as the basis for re-ranking subsequent result pages for the query. To further solve the word mismatch problem between queries and items, we proposed an end-to-end context-aware embedding model which can capture long-term and short-term context dependencies. Our experimental results on the datasets collected from the search log of a commercial product search engine show that short-term context leads to much better performance compared with long-term and no context. Our results also show thatmore »our proposed model is more effective than word-based context-aware models.« less
  4. Time series forecasting with additional spatial information has attracted a tremendous amount of attention in recent research, due to its importance in various real-world applications on social studies, such as conflict prediction and pandemic forecasting. Conventional machine learning methods either consider temporal dependencies only, or treat spatial and temporal relations as two separate autoregressive models, namely, space-time autoregressive models. Such methods suffer when it comes to long-term forecasting or predictions for large-scale areas, due to the high nonlinearity and complexity of spatio-temporal data. In this paper, we propose to address these challenges using spatio-temporal graph neural networks. Empirical results on Violence Early Warning System (ViEWS) dataset and U.S. Covid-19 dataset indicate that our method significantly improved performance over the baseline approaches.
  5. Deep neural network models, especially Long Short Term Memory (LSTM), have shown great success in analyzing Electronic Health Records (EHRs) due to their ability to capture temporal dependencies in time series data. When applying the deep learning models to EHRs, we are generally confronted with two major challenges: high rate of missingness and time irregularity. Motivated by the original PACIFIER framework which utilized matrix decomposition for data imputation, we applied and further extended it by including three components: forecasting future events, a time-aware mechanism, and a subgroup basis approach. We evaluated the proposed framework with real-world EHRs which consists of 52,919 visits and 4,224,567 events on a task of early prediction of septic shock. We compared our work against multiple baselines including the original PACIFIER using both LSTM and Time-aware LSTM (T-LSTM). Experimental results showed that our proposed framework significantly outperformed all competitive baseline approaches. More importantly, the extracted interpretative latent patterns from subgroups could shed some lights for clinicians to discover the progression of septic shock patients.