curaited.dev
  1. Home
  2. » 2025-07-31
  3. » Deep Learning

Deep Learning Advancements: From Privacy to Wearable Tech

Recent progress in deep learning spans multiple domains. One area focuses on privacy, with exploration of Fully Homomorphic Encryption (FHE) and its applications. Another highlight is the development and review of SensorLM, a family of sensor-language models that translate wearable sensor data into human-readable language, and LSM-2, which learns from incomplete wearable sensor data. In addition, Google Research is applying deep learning to contactless heart rate monitoring using ultra-wideband radar and introducing Graph Foundation Models (GFM) to enhance machine learning using interconnected relational tables. Finally, improvements to mobile transcription have been made, with SpeechCompass using multi-microphone localization for speaker diarization.

calendar_today 2025-07-18 attribution www.jeremykun.com/

Frequently Asked Questions about FHE

This blog post addresses frequently asked questions about Fully Homomorphic Encryption (FHE). FHE allows computation on encrypted data without decryption, ensuring privacy. It uses cryptographic schemes where operations on ciphertexts yield results corresponding to operations on the underlying plaintext. The author discusses the practicality of FHE in various applications like facial recognition, its quantum-resistant security, and its potential in scenarios where privacy is paramount due to legal requirements, insider risks, or the need for novel services. The article also compares FHE with other privacy-enhancing technologies, like SGX, TEE, and CVMs.
Good summary?
⊂
calendar_today 2025-07-17 attribution research.google/blog/

Measuring heart rate with consumer ultra-wideband radar

Google Research explores using ultra-wideband (UWB) radar, common in mobile phones, for contactless heart rate monitoring. By applying transfer learning, a deep learning model trained on FMCW radar data was adapted to UWB radar, achieving a mean absolute error of 4.1 bpm. This demonstrates the potential for leveraging existing hardware in consumer devices for health monitoring, paving the way for wider adoption of contactless heart rate measurement in everyday settings and continuous health tracking using mobile devices.
Good summary?
⊂
calendar_today 2025-07-10 attribution research.google/blog/

Graph foundation models for relational data

Google introduces Graph Foundation Models (GFM) to leverage interconnected relational tables for improved machine learning. By transforming tables into heterogeneous graphs, GFM captures crucial signals from the graph structure, outperforming traditional tabular methods. A key finding is that models trained on how features interact with each other in diverse tasks leads to better generalization. GFM demonstrates significant performance boosts in internal classification tasks like spam detection in ads, achieving substantial gains compared to single-table baselines, marking a significant step forward in graph learning and tabular ML.
Good summary?
⊂
calendar_today 2025-07-02 attribution research.google/blog/

Making group conversations more accessible with sound localization

Existing mobile transcription apps struggle to differentiate speakers in group conversations. Google's SpeechCompass enhances mobile captioning using multi-microphone localization for speaker diarization and directional guidance. It employs a TDOA-based algorithm with GCC-PHAT to accurately localize sound, offering visual cues like color-coded text and directional arrows. User studies showed SpeechCompass accurately localizes sound and improves diarization, making group conversations more accessible. The approach has lower computational costs, reduces latency, and enhances privacy preservation compared to ML approaches.
Good summary?
⊂
calendar_today 2025-07-22 attribution research.google/blog/

LSM-2: Learning from incomplete wearable sensor data

The blog post introduces LSM-2 with Adaptive and Inherited Masking (AIM), a novel self-supervised learning approach that learns directly from incomplete wearable sensor data. AIM addresses the challenge of missing data in wearable sensor streams by treating missingness as a natural artifact, improving upon previous models by reducing reliance on imputation or data removal. LSM-2 demonstrates strong performance across classification, regression, and generative tasks, proving more robust and scalable than its predecessor, especially when handling sensor failures or incomplete data.
Good summary?
⊂