Machine Learning in IoT Applications
summary
Argues that pushing ML to gateways cuts latency and bandwidth and boosts uptime: predictive maintenance on shop-floor sensors moves from scheduled checks to fewer, planned repairs and increases tool uptime; cites US healthcare spending $2.6T with $600B wasted and a UnitedHealthcare pilot that cut readmissions using Bluetooth monitoring. Describes stack and tools: Intel Clear Containers for near–bare-metal secure edge services, Intel Curie pattern engine (128 parallel units, 32,000 categories in 3-class inference), plus Spark Streaming, Kafka, InfluxDB, TensorFlow/Caffe and PMML model deployment.
AI-written summary · about
chapters (22)
- 0:00Intro
- 2:57A brief intro about me: Yashesh Shroff
- 5:49Smart Manufacturing and monitoring
- 13:35Make buildings and cities smarter
- 15:57loT - ML: High Level Platform Architecture
- 21:50Time-series data is not domain-specific
- 22:08Predictive Analysis
- 24:24Data analytics requirements for loT applications
- 25:59Time-series models
- 27:18Time-series analytics workflow
- 28:25Intel Deep Learning SDK
- 29:15Intel Curie
- 30:46Can Machines Really Learn?
- 31:52ML Model Performance
- 32:31Machine Learning Overview
- 33:07Regression Modeling Techniques
- 33:28Clustering
- 33:59How to train your Neural Network
- 35:44Handwriting recognition
- 36:20STANDARDIZE THE IOT INDUSTRY
- 41:40BigDL for Apache Spark
- 45:35Decision Trees Classification
related talks
20mHow ML is Powering the Future of IoT
01h 01mML and the IoT: Living on the Edge
32mExtending machine learning to industrial IoT applications at the edge with AWS
49mEnable IoT with Edge Computing and Machine Learning
23mKeynote: Intelligent Internet of Things: Start Analyzing Your Global Device Data for Real-Time
54mSmarter Everything with ML & the IoT
29mML-Powered IoT: From Warehouse Data to Prod Intelligence in Real Time
41mInfrastructure for ML Applications