Four short links: 9 July 2018
DNA Neural Nets, Ethics of AI, Oblivious Search, Physical Products
- Scaling up Molecular Pattern Recognition with DNA-based Winner-Take-All Neural Networks (Nature) — they use two molecular bio techniques (DNA-strand-displacement and a “seesaw DNA gate motif”) to implement a winner-takes-all type of neural network…with a soup of DNA. It recognizes digits from a 10×10 grid of pixels. The network successfully classified test patterns with up to 30 of the 100 bits flipped relative to the digit patterns “remembered” during training, suggesting that molecular circuits can robustly accomplish the sophisticated task of classifying highly complex and noisy information on the basis of similarity to a memory. (via The Next Web)
- The Ethics and Governance of Artificial Intelligence (MIT) — video from three classes is online.
- Oblix: An Efficient Oblivious Search Index — the new word I learned was “oblivious,” meaning that the actions of the algorithm over encrypted data do not reveal which (encrypted) documents match the keyword being searched for. Paper a Day makes sense of their work.
- Sonos One and Amazon Alexa Teardowns — using the physical product engineering to relate the smart speaker market positions of Sonos and Amazon. I work with mechanical and electrical engineers, and the invisible degrees of complexity in physical products always amazes me.