TimePerceiver: An Encoder-Decoder Framework for Generalized Time-Series Forecasting
About
In machine learning, effective modeling requires a holistic consideration of how to encode inputs, make predictions (i.e., decoding), and train the model. However, in time-series forecasting, prior work has predominantly focused on encoder design, often treating prediction and training as separate or secondary concerns. In this paper, we propose TimePerceiver, a unified encoder-decoder forecasting framework that is tightly aligned with an effective training strategy. To be specific, we first generalize the forecasting task to include diverse temporal prediction objectives such as extrapolation, interpolation, and imputation. Since this generalization requires handling input and target segments that are arbitrarily positioned along the temporal axis, we design a novel encoder-decoder architecture that can flexibly perceive and adapt to these varying positions. For encoding, we introduce a set of latent bottleneck representations that can interact with all input segments to jointly capture temporal and cross-channel dependencies. For decoding, we leverage learnable queries corresponding to target timestamps to effectively retrieve relevant information. Extensive experiments demonstrate that our framework consistently and significantly outperforms prior state-of-the-art baselines across a wide range of benchmark datasets. The code is available at https://github.com/efficient-learning-lab/TimePerceiver.
Key insights
What forecasting problem does TimePerceiver solve?
It generalizes time-series forecasting to jointly handle extrapolation, interpolation, and imputation. The encoder compresses arbitrary input patches into latent bottleneck tokens, and a query-based decoder retrieves target timestamps, enabling single end-to-end training without separate pretraining.
How does it differ from prior forecasting approaches?
Unlike prior work focused on encoder design with simple linear decoders, it unifies architecture and training. Across 80 settings on 8 datasets, it achieves 55 best and 17 second-best scores, with the best average rank of 1.375 in MSE versus 10 baselines.
What are the practical uses of these findings?
The framework suits multivariate forecasting benchmarks on regular-interval data such as weather, electricity, traffic, and solar domains. The authors believe it can extend to reconstructing unobserved data in healthcare, finance, and environmental monitoring, though no production deployment is documented.
What limitations and future work do the authors note?
The framework is evaluated only on regular-interval multivariate benchmarks. The authors identify extending it to irregular or event-based time series as an important future direction, noting their latent bottleneck design could support this but has not been tested.