IEEE Signal Processing Society

Most CNN-based biomedical reconstruction methods rely on trainable regularizers or PnP denoisers—but both need strict stability constraints that limit model expressivity. Dr. Unser will show how making activation functions trainable (with controlled slopes) can break this bottleneck and open the door to more powerful reconstruction networks.

⚡Register at https://hubs.la/Q043z4tG0

Join the world’s largest technical conference dedicated to signal processing—built on 50 years of innovation, breakthroughs, and global research.

Hear from industry leaders and pioneering academics, connect with experts from around the world, explore programs designed for students and young professionals, and so much more.

⚡Register at https://hubs.la/Q042hq830.

Look Who’s Talking 👀🗣️

Spoken Language Models are leveling up. This talk dives into how today’s text‑based LLMs are learning to hear and speak—processing and generating natural speech without losing any of their powerhouse abilities. The future of voice‑native AI is getting loud.

🔗Register at https://hubs.la/Q041QWZy0

Unlocking 6G Potential!

ELAA systems are key to major 6G performance improvements, but we need to tackle the challenges of computational complexity and interconnection cost.

The solution? Distributed signal processing! It offers a scalable pathway by moving baseband tasks—like channel estimation and beamforming—from centralized processing to cooperative designs with local computing and limited information exchange.

🔗Read the article at https://hubs.la/Q041whRv0.

Quantum tech is taking a big leap. Dr. Liu introduces a new mixed analog‑digital approach to quantum signal processing that bridges qubits with continuous‑variable systems. 🚀

🔗Register at https://hubs.la/Q03_Jf5v0.