1. Home
  2. » 2025-10-31
  3. » Deep Learning

Revolutionizing Intelligence: New Frontiers in Edge AI and Cancer Research

Recent advancements in artificial intelligence are significantly expanding its applications, particularly through original research and development. Pioneering efforts are being made in both ubiquitous personal AI and precision healthcare. For personal AI, a new full-stack, open-source platform, including a RISC-V based NPU, has been introduced to enable powerful, always-on machine learning and small transformer models on battery-constrained edge devices like wearables. This innovation addresses challenges in compute and privacy with hardware-enforced security, facilitating features such as ambient sensing and real-time translation. Concurrently, in medical research, an AI-powered tool leveraging convolutional neural networks has been developed to identify cancer-related mutations with exceptional accuracy. This flexible machine learning model, designed to outperform existing methods, aims to accelerate cancer research, improve treatment decisions, and advance precision medicine by pinpointing somatic variants in various cancer types.

calendar_today 2025-10-15 attribution research.google/blog/

Coral NPU: A full-stack platform for Edge AI

The future of AI is personal, but embedding powerful models on battery-constrained edge devices presents significant challenges in compute, fragmentation, and privacy. Google introduces Coral NPU, a full-stack, open-source platform with an AI-first architecture designed for ultra-low-power, always-on edge AI. This RISC-V based NPU and its comprehensive toolchain enable efficient deployment of complex ML and small transformer models on wearables and IoT devices. It prioritizes hardware-enforced security and aims to foster a unified, open ecosystem for next-generation personal AI, enabling features like ambient sensing and real-time translation.
Good summary?
calendar_today 2025-10-16 attribution research.google/blog/

Using AI to identify genetic variants in tumors with DeepSomatic

Discovering the genetic underpinnings of cancer is crucial for effective treatment, and Google Research introduces DeepSomatic, an AI-powered tool designed to revolutionize how we identify cancer-related mutations. This flexible machine learning model leverages convolutional neural networks to pinpoint somatic variants in tumors with unprecedented accuracy. DeepSomatic outperforms existing methods, particularly for insertions and deletions, across major sequencing platforms and various sample types, including challenging FFPE and tumor-only cases. Openly available, it aims to accelerate cancer research and advance precision medicine by identifying new and known cancer variants in breast, lung, glioblastoma, and leukemia, aiding treatment decisions and drug discovery.
Good summary?