dotAI 2017
who it's for
deep learning engineers, computer vision engineers, machine learning engineers, mobile Android developers, data engineers, ml infrastructure engineers, product-focused data scientists
topics
recommender system embeddings; triplet loss siamese networks; session embedding for recommendations; TensorFlow on Android; Inception v3 transfer learning; model quantization and pruning; satellite imagery deep learning; continuous training and versioning; gpu kernel optimization and tiling; reinforcement learning in JavaScript; image feature extraction with NMF
| title | author | length |
|---|---|---|
| Neural Networks for Recommender Systems | Olivier Grisel | 21m |
Claims that neural nets reproduce and extend classic methods: matrix factorization = dot product of embeddings and replacing the dot with an MLP yields more… | ||
| On-device machine learning: TensorFlow on Android | Yufeng Guo | 19m |
Demo shows a real Android app doing on‑device inference from the camera, correctly classifying specific candy packages while offline (airplane mode). Workflow… | ||
| Deep learning for Space imagery applications | Mathias Ortner | 18m |
Applies deep learning to satellite imagery sized 40,000×400,000 pixels and delivers concrete results: cloud detection cuts error rate by 4× versus prior state… | ||
| AI Trends in Intelligent Transport Systems | Laurent Smadja | 19m |
History of vision models leads to concrete claims: Navya runs classical image‑processing for road marking and obstacle detection and is porting deep networks… | ||
| Optimizing for GPUs | Arnaud Bergeron | 19m |
Concrete measurements and recipes: kernels run in warps of 32 or 64 so choose block sizes that avoid idle threads; grid sizing matters — Titan X favors about… | ||
| Alpha2048 | Lian Li | 20m |
Built a browser prototype called Alpha2048 that uses JavaScript libraries synaptic and reinforced.js with Q‑learning. Data collection starts from serialized… | ||
| Pragmatic Deep Learning for image labelling | Pierre Gutierrez | 20m |
An image pipeline labels hotel photos using transfer learning from pre-trained models in the Caffe model zoo and public datasets (Places and SUN), then… | ||
| nothing matches | ||