ICSPIS 2025 draws researchers from 12 countries: practical advances in AI and signal processing
The 11th International Conference on Signal Processing and Intelligent Systems (ICSPIS 2025) convened on December 24-25 at Mazandaran University of Science and Technology, held in person and online. Researchers from Iran and 12 countries-Spain, Italy, Iceland, Australia, Mexico, China, India, England, Algeria, Denmark, Turkey, and Azerbaijan-took part.
The program mixed educational workshops with focused technical sessions. Out of 239 submissions, 125 papers were accepted (about 52 percent), with 90 percent presented in English.
What the community focused on
Discussion centered on practical AI, deep learning, and advanced signal processing applications-areas where reproducibility, dataset quality, and deployment constraints matter. The format encouraged direct knowledge exchange and quick alignment on shared research problems. For resources on research workflows, reproducibility, and tools, see AI Research - Courses & Tools.
Conference tracks
- Track 1: Signal Processing; Medical Image Processing; Audio and Speech Processing; Image and Video Processing; Remote Sensing
- Track 2: Pattern Recognition; Machine Learning; Data Mining; Robotics; Soft Computing
- Track 3: Smart Computer Networks; Smart Grids; Internet of Things (IoT); Industrial Automation - practical deployment topics covered under AI for Operations
Why this matters for researchers
If you work on models for perception, compression, or control, this conference mapped where the field is actually moving. The mix of application-heavy tracks (medical imaging, speech, remote sensing) and systems tracks (IoT, smart grids, automation) points to demand for reliable, efficient methods that perform under resource and data constraints.
Given the acceptance stats, English-language submissions dominate. For international reach and citations, preparing full English packages-paper, slides, code, and data-remains the practical choice.
Iran's current position in AI research and readiness
The country's standing in AI-related scientific output has moved from 33 to 30, indicating steady progress in publication impact. Regionally, its ranking sits between 14 and 17, with officials signaling a push on infrastructure, shared AI platforms, and assistant technologies.
On government readiness for AI in public services, Iran ranks 91st out of 188 in the latest Oxford Insights Government AI Readiness Index, up three places from 2023. The strongest pillar is Data and Infrastructure at 66.29 (up from 55.88), with indicator signals in infrastructure (70), data availability (43), and data representativeness (121).
For broader context on scientific output trends across fields, see Nature Index.
Actionable takeaways for your lab
- Prioritize end-to-end reproducibility: publish code, trained weights, and small, clean test sets where licensing allows.
- Target cross-track problems-e.g., remote sensing models deployed over constrained networks, or medical imaging pipelines optimized for edge devices.
- Build bilingual assets only if your funders require it; otherwise keep the full stack in English to reduce friction for collaborators.
- Plan follow-ups: convert workshop demos into preprints, and propose multi-institution benchmarks tied to the three tracks.
If you're formalizing team upskilling around these topic areas, you can scan curated AI course lists by skill here: Complete AI Training - Courses by Skill, or follow the AI Learning Path for Data Scientists for end-to-end pipeline training.
Your membership also unlocks:
