Shyamal Chandra

This paper is essential a neural key-value-like store with probably (at first glance) no consideration (hopefully) about the previous states that it has gone to in the past in heuristic search.  They are concerned with the learning acquisition rate in this scenario.  What about facets like energy, scalability, time, asymptotic runtime, cost in terms of money, space in terms of cubic feet, and correlation to the human mind in terms of number of examples?

This is moving towards general problem solvers (GPS) but this is sans planning like AlphaGo with only a layer of a mixture of experts.  Pathnet does do transfer learning by evolved layers of channels but if you combine the two, you might get surprising results similar to the Minsky’s way of thinking in Emotion Machine where emotion is only directed search.  One could imagine many possibilities for this avenue with heuristic search and path planning in addition to addition, subtraction, combination, and splitting of layers to form the ultimate, plastic and malleable AI.  Furthermore, there could be elements of CoDeepNEAT that could be added to the mix.

In principle, You can beat Captcha (or any Invisible Captcha or reCaptcha) with Mechanical Turk by outsourcing the VNC screen to low-wage automatons, not n-shot Deep learning or even one-shot BPL.  One is making the problem harder than it should be in this scenario.

Lennon (Rock-n-roll) +
A-ha (New Wave) +
Chicago (Power Ballet) +
Eagles (70’s Rock)
= Love Away!

Yann LeCun: How would you compare this offering to the DGX-1 from Nvidia, Google TensorFlow ASICs available on the G-Cloud, TerraDeep, D-Wave 1000-Qbit QPU [with quantum DL], and other deep learning-compatible hardware that can both train and test without using Spark or Map/Reduce? What about cost, time to perform a cross-validation, space (cubic feet), power, and scalability?  What about a pure-GPU computer without any CPU?

Fake History +
Fake MSM news funded by the 1% +
Fake Promises by politicians (controlled by lobbyist) +
Fake Demos of Singularity (promise strong AI & deliver weak AI) =
Fake World made to rule by the 1%!

• Why doesn’t Bernie Sanders advocate using the tax money from the 1% to fund universal exercise programs for the 99%?
• Why doesn’t Bernie Sanders advocate using the tax money from the 1% to fund universal education/startup programs for social mobility for the 99%?

All this talk about robin hood-esqe tax breaks doesn’t result in better financial utilization of monetary funds by the 99% and is quite short-sided because the people will eventually not be able to climb the social ladder of financial and job independence with control by large corporations and be stalled in their quest to become the master of their destinies.  Startup incubators could be funded to give money to small businesses or part-time work for off-the-job entrepreneurship.

Universal education with checks-and-balances and full transparency with levels of privacy would provide a better route to get to not work for the big man but rather build a small business that results in higher cut of the profits of the venture.  The USA needs to understand that simply shifting money from one group of people to another will result in tons of corruption, bad financial spending habits, and long-term understanding of the betterment for consumers that would like competition and choices.

Simple exercise can prevent complex medical burden on the American taxpayers.

1. Look ahead of links/alter links (0 citations according to Google Scholar)
2. Architecture for user- and context-specific prefetching and caching of information on portable devices (5 citations according to Google Scholar)
3. Data management in social networks (2 citations)
4. Search guided by location and context (0 citations)
5. Search engine that identifies and uses social networks in communications, retrieval, and electronic commerce (0 citations)

Total citations: 7 on single-author patents

Today, let’s look at the top 8 cited patent portfolio that included Steve Jobs as one of the many co-authors during his two tenures are Apple Computer:

1. Computing device (2002) — Design patent on the MacBook (prior art: Dynabook: 1972 – Alan Kay)
2. Electronic device (2005) — Design patent on the iPad (prior art: Telautograph: 1888 – Elisha Gray)
3. Media device (2009) — Design patent on the iPod Touch (prior art: IXI: 1979 – Kane Kramer)
4. Electronic device with GUI (2010) — Design patent on the iPhone (prior art: Touch Screen: 1965 – E.A. Johnson)
5. Media Device (2005) — Design patent on the iPod nano (prior art: Click Wheel: Synaptics — unknown – co-founder: Carver Mead)
6. Media Device (2004) — Design patent on the iPod shuffle (prior art: popular MP3 player: MPMan: 1998 — SaeHan Information Systems)
7. Portable display device (2010) — Design patent on the iPad (prior art: portable tablet PC: GRiDPad: 1989 — Samsung Corporation)
8. Portable Computer (2009) — Design patent on the MacBook (prior art: portable laptop PC: Osborne 1: 1981 — Osborne Corporation)