Deep-Learning image recognition is a hot-topic. The billion dollar thought is to create a "Google" of image search (or a mesh-search engine for 3D printing and animation), but that requires rather high search fidelity. A lower-fidelity approach is to use key information provided in selfies - specifically the identity of the individual in the … [Read more...]
Depth-Categorizing GPU-Accelerated Deep Neural Networks Perform Fast Semantic Segmentation of RGB-D Scenes
Interesting for computer vision and animation, the paper by Nico Höft, Hannes Schulz, and Sven Behnke, "Fast Semantic Segmentation of RGB-D Scenes with GPU-Accelerated Deep Neural Networks" categorizes the surface to which each pixel in an image belongs. Semantic scene segmentation is a major challenge on the way to functional computer vision systems that can separately label … [Read more...]
Deep-learning Webinar Demonstrates Handwriting Recognition and Efforts to Teach Drone to Fly Down a Wooded Path
Deep-learning is a computational expensive but rewarding method to solve many complex pattern recognition problems. The recent NVIDIA webinar by Dan Claudiu Cireșan, Senior Researcher at the Dalle Molle Institute for Artificial Intelligence (IDSIA) in Switzerland highlighted some of the capabilities of deep-learning for image recognition problems such as handwriting recognition … [Read more...]
IBM TrueNorth a “Bee Brain” on a SyNAPSE Chip That Uses 70 mW
IBM unveiled the first neurosynaptic computer chip on August 7th that implements one million programmable neurons, 256 million programmable synapses and 46 billion synaptic operations per second per watt. The IBM announcement, published in Science in collaboration with Cornell Tech, is a step towards bringing cognitive computers to society.At 5.4 billion transistors, this fully … [Read more...]
GaussianFace: Computers Claimed to Beat Humans in Recognizing Faces
In a human vs. computer test on 13k photos of 6k public figures, the GaussianFace project claims to identify human faces better than humans (97% human accuracy vs. 98% computer accuracy). The authors claim their model can adapt automatically to complex data distributions, and therefore can well capture complex face variations inherent in multiple sources. The reporters at The … [Read more...]
Deep-learning Teaching Code Achieves 13 PF/s on the ORNL Titan Supercomputer
The deep-learning teaching code described in my book, "CUDA Application Design and Development" [Chapters 2, 3, and 9] plus online tutorials achieved 13 PF/s average sustained performance using 16,384 GPUs on the Oakridge Titan supercomputer. Full source code for my teaching code can be found on github in the farbopt directory. Nicole Hemsoth at HPCwire noted these CUDA … [Read more...]