- Washers and Screws (YouTube) — this chap is making his own clock from scratch, and here he is making his own washers and screws. Sometimes another person’s obsession can be calming. (via Greg Sadetsky)
- ROScon 2015 Recap with Videos (Robohub) — Shuttleworth suggests that robotics developers really need two things at this point: a robust Internet of Things infrastructure, followed by the addition of dynamic mobility that robots represent. However, software is a much more realistic business proposition for a robotics startup, especially if you leverage open source to create a developer community around your product and let others innovate through what you’ve built.
- Getting Deep Speech to Work in Mandarin (Baidu SVAIL) — TIL that some of the preprocessing traditionally used in speech-to-text systems throws away pitch information necessary to decode tonal languages like Mandarin. Deep Speech doesn’t use specialized features like MFCCs. We train directly from the spectrogram of the input audio signal. The spectrogram is a fairly general representation of an audio signal. The neural network is able to learn directly which information is relevant from the input, so we didn’t need to change anything about the features to move from English speech recognition to Mandarin speech recognition. Their model works better than humans at decoding short text such as queries.
- Sequencing Genomes of All Known Kakapo — TIL there’s a project to sequence genomes of 10,000 bird species and that there’s this crowdfunded science project to sequence the kakapo genome. There are only 125 left, and conservationists expect to use the sequenced genomes to ensure rare genes are preserved. Every genome in this species could be sequenced … I’m boggling. (via Duke)
"ai" entries
Four short links: 16 February 2016
Full-on Maker, Robot Recap, Decoding Mandarin, and Sequencing Birds
Four short links: 9 February 2016
Collaborative Mario Agents, ElasticSearch at Scale, Anomaly Detection, Robotics Experiment
- Social Intelligence in Mario Bros (YouTube) — collaborative agents built by cognitive AI researchers … they have drives, communicate, learn from each other, and solve problems. Oh, and the agents are Mario, Luigi, Yoshi, and Toad within a Super Mario Brothers clone. No code or papers about it on the research group’s website yet, just a YouTube video and a press release on the university’s website, so appropriately adjust your priors for imminent world destruction at the hands of a rampaging super-AI. (via gizmag)
- How we Monitor and Run ElasticSearch at Scale (SignalFx) — sweet detail on metrics, dashboards, and alerting.
- Simple Anomaly Detection for Weekly Patterns — Rule-based heuristics do not scale and do not adapt easily, especially if we have thousands of alarms to set up. Some statistical approach is needed that is generic enough to handle many different metric behaviours.
- How to Design a Robotics Experiment (Robohub) — although there are many good experimental scientists in the robotic community, there has not been uniformly good experimental work and reporting within the community as a whole. This has advice such as “the five components of a well-designed experiment.”
Four short links: 4 February 2016
Shmoocon Video, Smart Watchstrap, Generalizing Learning, and Dataflow vs Spark
- Shmoocon 2016 Videos (Internet Archive) — videos of the talks from an astonishingly good security conference.
- TipTalk — Samsung watchstrap that is the smart device … put your finger in your ear to hear the call. You had me at put my finger in my ear. (via WaPo)
- Ecorithms — Leslie Valiant at Harvard broadened the concept of an algorithm into an “ecorithm,” which is a learning algorithm that “runs” on any system capable of interacting with its physical environment. Algorithms apply to computational systems, but ecorithms can apply to biological organisms or entire species. The concept draws a computational equivalence between the way that individuals learn and the way that entire ecosystems evolve. In both cases, ecorithms describe adaptive behavior in a mechanistic way.
- Dataflow/Beam vs Spark (Google Cloud) — To highlight the distinguishing features of the Dataflow model, we’ll be comparing code side-by-side with Spark code snippets. Spark has had a huge and positive impact on the industry thanks to doing a number of things much better than other systems had done before. But Dataflow holds distinct advantages in programming model flexibility, power, and expressiveness, particularly in the out-of-order processing and real-time session management arenas.
Four short links: 2 February 2016
Fourth Industrial Revolution, Agent System, Evidence-Based Programming, and Deep Learning Service
- This is Not the Fourth Industrial Revolution (Slate) — the phrase “the fourth Industrial Revolution” has been around for more than 75 years. It first came into popular use in 1940.
- Huginn — MIT-licensed system for building agents that perform automated tasks for you online. They can read the Web, watch for events, and take actions on your behalf. Huginn’s Agents create and consume events, propagating them along a directed graph. Think of it as a hackable Yahoo! Pipes plus IFTTT on your own server.
- Evidence-Oriented Programming — design programming language syntax and features based on what research shows works. They tested Perl and Java, found apparently not detectably easier to use for novices than a language that my student at the time, Susanna Kiwala (formerly Siebert), created by essentially rolling dice and picking (ridiculous) symbols at random.
- Deep Detect — open source deep learning service.
Four short links: 1 February 2016
Curation & Search, Developer Tenure, AI/IA History, and Catapulting Drones
- Curation & Search — (Twitter) All curation grows until it requires search. All search grows until it requires curation.—Benedict Evans. (via Lists are the New Search)
- Average Developer Tenure (Seattle Times) — The average tenure of a developer in Silicon Valley is nine months at a single company. In Seattle, that length is closer to two years. (via Rands)
- An Interview with John Markoff (Robohub) — the interview will give you a flavour of his book, Machines of Loving Grace, a sweet history of AI told through the stories of the people who pioneered and now shape the field.
- Catapult Drone Launch (YouTube) — utterly nuts. That’s an SUV off its rear wheels! (via IEEE)
Four short links: 28 January 2016
Augmented Intelligence, Social Network Limits, Microsoft Research, and Google's Go
- Chimera (Paper a Day) — the authors summarise six main lessons learned while building Chimera: (1) Things break down at large scale; (2) Both learning and hand-crafted rules are critical; (3) Crowdsourcing is critical, but must be closely monitored; (4) Crowdsourcing must be coupled with in-house analysts and developers; (5) Outsourcing does not work at a very large scale; (6) Hybrid human-machine systems are here to stay.
- Do Online Social Media Remove Constraints That Limit the Size of Offline Social Networks? (Royal Society) — paper by Robin Dunbar. Answer: The data show that the size and range of online egocentric social networks, indexed as the number of Facebook friends, is similar to that of offline face-to-face networks.
- Microsoft Embedding Research — To break down the walls between its research group and the rest of the company, Microsoft reassigned about half of its more than 1,000 research staff in September 2014 to a new group called MSR NExT. Its focus is on projects with greater impact to the company rather than pure research. Meanwhile, the other half of Microsoft Research is getting pushed to find more significant ways it can contribute to the company’s products. The challenge is how to avoid short-term thinking from your research team. For instance, Facebook assigns some staff to focus on long-term research, and Google’s DeepMind group in London conducts pure AI research without immediate commercial considerations.
- Google’s Go-Playing AI — The key to AlphaGo is reducing the enormous search space to something more manageable. To do this, it combines a state-of-the-art tree search with two deep neural networks, each of which contains many layers with millions of neuron-like connections. One neural network, the “policy network,” predicts the next move, and is used to narrow the search to consider only the moves most likely to lead to a win. The other neural network, the “value network,” is then used to reduce the depth of the search tree — estimating the winner in each position in place of searching all the way to the end of the game.
Four short links: 26 January 2016
Inequality, Conversational Commerce, Minsky Lectures, and Trust vs Transparency
- What Paul Graham is Missing About Inequality (Tim O’Reilly) — When a startup doesn’t have an underlying business model that will eventually produce real revenues and profits, and the only way for its founders to get rich is to sell to another company or to investors, you have to ask yourself whether that startup is really just a financial instrument, not that dissimilar to the CDOs of the 2008 financial crisis — a way of extracting value from the economy without actually creating it.
- 2016 The Year of Conversational Commerce (Chris Messina) — I really hope that these conversations with companies are better than the state-of-the-art delights of “press 5 to replay” phone hell.
- Society of Mind (MIT) — Marvin Minsky’s course, with lectures.
-
Trust vs Transparency (PDF) — explanation facilities
can potentially drop both a user’s confidence and make the process of search more stressful. Aka “few takers for sausage factory tours.” (via ACM Queue)
Four short links: 20 January 2016
Rules-Based Distributed Code, Open Source Face Recognition, Simulation w/Emoji, and Berkeley's AI Materials
- Experience with Rules-Based Programming for Distributed Concurrent Fault-Tolerant Code (A Paper a Day) — To demonstrate applicability outside of the RAMCloud system, the team also re-wrote the Hadoop Map-Reduce job scheduler (which uses a traditional event-based state machine approach) using rules. The original code has three state machines containing 34 states with 163 different transitions, about 2,250 lines of code in total. The rules-based re-implementation required 19 rules in 3 tasks with a total of 117 lines of code and comments. Rules-based systems are powerful and underused.
- OpenFace — open source face recognition software using deep neural networks.
- Simulating the World in Emoji — fun simulation environment in the browser.
- Berkeley’s Intro-to-AI Materials — We designed these projects with three goals in mind. The projects allow students to visualize the results of the techniques they implement. They also contain code examples and clear directions, but do not force students to wade through undue amounts of scaffolding. Finally, Pac-Man provides a challenging problem environment that demands creative solutions; real-world AI problems are challenging, and Pac-Man is, too.
Is 2016 the year you let robots manage your money?
The O’Reilly Data Show podcast: Vasant Dhar on the race to build “big data machines” in financial investing.
Subscribe to the O’Reilly Data Show Podcast to explore the opportunities and techniques driving big data and data science.
In this episode of the O’Reilly Data Show, I sat down with Vasant Dhar, a professor at the Stern School of Business and Center for Data Science at NYU, founder of SCT Capital Management, and editor-in-chief of the Big Data Journal (full disclosure: I’m a member of the editorial board). We talked about the early days of AI and data mining, and recent applications of data science to financial investing and other domains.
Dhar’s first steps in applying machine learning to finance
I joke with people, I say, ‘When I first started looking at finance, the only thing I knew was that prices go up and down.’ It was only when I actually went to Morgan Stanley and took time off from academia that I learned about finance and financial markets. … What I really did in that initial experiment is I took all the trades, I appended them with information about the state of the market at the time, and then I cranked it through a genetic algorithm and a tree induction algorithm. … When I took it to the meeting, it generated a lot of really interesting discussion. … Of course, it took several months before we actually finally found the reasons for why I was observing what I was observing.
Four short links: 30 December 2015
Bitcoin Patents, Wall-Climbing Robot, English 2 Code, and Decoding USB
- Bank of America Loading up on Bitcoin Patents — The wide-ranging patents cover everything from a “cryptocurrency transaction payment system” which would let users make transactions using cryptocurrency, to risk detection, storing cryptocurrencies offline, and using the blockchain to measure fraudulent activity.
- Vertigo: A Wall-Climbing Robot (Disney Research) — watch the video. YOW! (via David Pescovitz)
- Synthesizing What I Mean — In this paper, we describe SWIM, a tool which suggests code snippets given API-related natural language queries.
- serialusb — this is how you decode USB protocols.