Fotor's Joint Research on "PMDformer" Accepted at ICLR 2026: A Leap Forward in Long-term Time Series Forecasting

Fotor [https://www.fotor.com], the flagship AI product of Everimaging, today announced that its latest joint research achievement in Long-term Time Series Forecasting (LTSF), PMDformer, has been accepted at ICLR 2026, a premier international conference in artificial intelligence.

Image: https://www.globalnewslines.com/uploads/2026/04/c3bf1e66bb1a5e36cabcacce394d604e.jpg

The paper, titled "PMDformer: Patch-Mean Decoupling Information Transformer for Long-term Forecasting [https://openreview.net/forum?id=rfJ41gK9Ct]," is the result of a collaborative research effort between Fotor and leading universities and research partners, successfully bridging the gap between academic achievements and real-world industrial practice.

By setting new benchmarks for forecasting accuracy and stability, PMDformer marks an important advance in the LTSF field. Beyond its academic significance, the technology is also expected to strengthen Fotor's global product ecosystem, delivering a more intelligent and cutting-edge image creation experience to hundreds of millions of users worldwide.

Tackling a Core Challenge in Forecasting

In real-world forecasting tasks such as traffic analysis and market trend prediction, highly volatile and noisy data has long posed a major challenge for conventional forecasting models. Traditional Transformer-based methods are often overly influenced by absolute values and outdated historical signals, which can limit their ability to capture meaningful long-term patterns.

To address these challenges, PMDformer introduces three core innovations: Patch-Mean Decoupling (PMD), which helps reduce bias from absolute magnitudes and better capture underlying fluctuation patterns; Proximal Variable Attention (PVA), which strengthens the model's focus on more relevant and timely variable interactions; and Trend Recovery Attention (TRA), which restores long-term trend information to improve both responsiveness and stability in forecasting.

Image: https://www.globalnewslines.com/uploads/2026/04/2cb7101954ae66ffd97a555cd3b03cf9.jpg

Together, these innovations enable PMDformer to deliver stronger forecasting accuracy and stability while also improving computational efficiency through lower memory usage.

From Research to Real-World Impact

As a leading global image processing and creation platform, Fotor is committed to landing the most cutting-edge AI technology in practical business scenarios. The birth of PMDformer injects powerful "prophetic" capabilities into Fotor's product evolution and commercial decision-making:

PMDformer is expected to strengthen Fotor's capabilities in product development, operational planning, and strategic decision-making. By improving the ability to detect emerging trends, shifts in user interest, and early signals of feature adoption, the research can support more informed decisions across Fotor's global platform.

Fotor consistently believes that top-tier AI research is not just for refreshing academic leaderboards, but more importantly, for resolving industry pain points and empowering the creativity of every user. The successful publication of PMDformer at ICLR 2026 marks another solid step forward for Fotor on the path of promoting the practical application of AI technology.

About Fotor

Fotor is a leading global photo editor and AI visual creation platform for everyday users, creators, marketers, and entrepreneurs worldwide. Dedicated to the practical application of advanced AI research, Fotor combines technical innovation with user-focused product design to make visual creation more accessible and efficient. Today, it empowers hundreds of millions of users worldwide to create, edit, and enhance visual content with ease.
Media Contact
Company Name: Fotor
Contact Person: Coran Huang
Email: Send Email [http://www.universalpressrelease.com/?pr=fotors-joint-research-on-pmdformer-accepted-at-iclr-2026-a-leap-forward-in-longterm-time-series-forecasting]
Country: United States
Website: https://www.fotor.com/

Legal Disclaimer: Information contained on this page is provided by an independent third-party content provider. GetNews makes no warranties or responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you are affiliated with this article or have any complaints or copyright issues related to this article and would like it to be removed, please contact retract@swscontact.com



This release was published on openPR.