Bits & Bytes
- Apple launched their Machine Learning Journal site to publish their research. The first article is on a technique for improving the realism of fake images using a technique similar to GANs.
- Google updated their 9 million large Open Images dataset to add some 2 million bounding boxes and several million more labels.
- OpenAI published research on a new approach to reinforcement learning called Proximal Policy Optimization (PPO). PPO aims to outperform the current state-of-the-art methods while being simpler to implement.
- In a paper published in Neuron, Google DeepMind founder Demis Hassabis and co-authors argue that understanding human intelligence is the key to creating artificial intelligence.
- An interesting discussion of some ways technical debt is accumulatedin machine learning projects.
- Harvard Business Review features a nice profile of Facebook’s Applied Machine Learning group.
- The IEEE Computer Vision and Pattern Recognition (CVPR) conference just ended. Best Paper winners were Densely Connected Convolutional Networks and Learning from Simulated and Unsupervised Images through Adversarial Training. Perhaps someone reading this would like to present one of these papers at an upcoming meetup?