Building Literate Machines
summary
Maluuba (now part of Microsoft) builds neural systems that read web pages and answer questions in real time; a live demo shows a Chrome extension parsing articles it never saw during training and returning concise answers. Models achieve near state‑of‑the‑art results on datasets like NewsQA and CNN QA, use deep nets for machine reading comprehension, and apply reinforcement learning to generate useful follow‑up questions. Claims include billions of queries and strong SOTA comparisons, with current limits in transfer learning, common‑sense reasoning, and multimodal grounding.
AI-written summary · about
chapters (8)
- 0:00Start
- 0:30Using Deep Learning To Really Understand Language
- 6:21Core Applications of Artificial Intelligence
- 11:10Limitations in Ai
- 13:03Ibm Watson
- 20:28Microsoft's Blog about Tackling Water Scarcity
- 21:39Conversational Interfaces
- 26:23What's Next
related talks
24mMachine Intelligence gets real
47mBuilding smarter apps with Machine Learning, from magic to reality
44mArtificial Intelligence and Machine Learning in the browser
35mIndustrial Applications of Machine Learning in Search
40mMachines are Learning
01h 03mIt's All About AI
57mBuilding smarter solutions with Machine Learning, from magic to reality
27mLive and Machine Learn