61 karma · joined April 6, 2012
Facebook is famous for cassandra for example.
They are tons of uses for them. A lot of them are evented and stored data analysis.
People create new ones because performance is even more necessary if your business runs on these things for many services.
If yes to all above then you can maybe turn that into a product. A great site for seeing what is available for frameworks is yeoman: http://yeoman.io/generators/ .
Something else to be aware of is if you develop something now you are most likely to see it's production use within 3-5 years of development and then you are most likely to see it's market either growing or not in 10 years. This is an interesting dichotomy because it means picking things that may be hard but boring, or fun depending on how you look at them.
I am not affiliated with any of these companies. I just think some of them are well-placed in the market.
-Devops (mostly containers and virtualization software, but also fast/easy build software, customizable configs that adapt better and searchable/informative apis)
-Devops massive ssl/managing keys/passwords deployment software (there are a few automated tools that may help with this but there is in general not too many companies currently handling it)
-Devops massive worker model software or software that helps automation in newer/interesting ways like stackengine (http://stackengine.com/blog/) just got $1 million in funding and provides a way to convert vagrant images to docker images and back again in a fast fashion
-Devops with companies that provide continuous integration at scale or preconfigured devops things that require too much learning to integrate etc, or alternatively lease integration tools. Also managed data worflow between many integrative models. https://jidoteki.com/
-Saas, PaaS, etc software as a service especially in devops looks to be going in several directions or so: 1)safe php applications using sandstorm sandbox and type-safe php llvm type stuff (integrating some of parse: https://parse.com/docs/php_guide which makes traveling from one php application to another seamless, and Hoa which makes dsls more manageable http://hoa-project.net/En/ ) 2)devops delivered through docker + automated binary releases 3)devops in nodejs that controls docker and other things or just low latency/fast productive js (dart could be a good useful bet) 4)determinable execution type programming mostly ocaml and haskell which are considered 90% correct and safer languages 5)clojure immutable data that can be shared between large amounts of web online applications
-safe low latency/low requirement distributed services
-large data crunching (think distributed versus not distributed, purpose of what it is doing, competition)
-the coming large amount of "internet of things" in the next few years will be tons of small robots because of finally cheap enough robotics that they will be buyable + good software markets from many different companies working together The internet of things opens a new dimension for SaaS products that interact seamlessly this is an idealistic idea for how to structure something like that as many multiple agents: http://eve.almende.com/ Also training an AI in a massive simulation with cloud/distributed tech with machine learning algorithms for real usage is a definite possibility
-privacy software (anonymous logins like facebook did, garbled circuits and homomorphic encryption, encrypted multiparty applications like auctions for bidding etc and cloud computing)
-artificial intelligence (not just "deep learning" on it's own like convnets which are doing good performance wise but lack overall structural use, but intelligent software that can solve problems using many other algorithmic components like bloom filters, markov chains, dynamic time warping etc that focuses on specific problem solving interactive on-work jobs, data cleaning, vision, speech recognition and NLP)As a side-note if someone was able to improve data cleaning they can reduce a lot of errors by a large % and is under-utilized at the moment. Although it is problematic that most active learning algorithms ask for a lot more quesitons: https://amplab.cs.berkeley.edu/publication/scaling-up-crowd-... . Crowd answering has been a better use for data cleaning.)
-Specifically Artificial Intelligence in vision through point cloud library needs to be more useful. Vision will need to be a very automated and big process and is not talked about as much on how hard it is to determine vision constraints like object detection and avoiding using edge detection etc. Simply learning to move is not the only problem associated with AI since vision is hard: http://pointclouds.org/
-Specifically Artifical Intelligence using networking to off-load some of the heavy load in computing lots of difficult/complicated tasks by automating large chunks of activity as computer signals this is known as macroprogramming: http://fiji.eecs.harvard.edu/Macroprogramming )
-3D Printing (I think metal 3d printing or harder material like construction from cement is more successful than regular plastic 3d printing as it can lead to more things getting done. 3d printing could be improved wholesale from some re-imaging by making it from adaptable molds rather than a dripple of material that moves (different concept than most 3d printing and may provide different results). Cheaper materials that are durable integration into actual electronics or support for holding computer chips in the printed models)
-Better safer faster payment gateways
-Anything security that helps solve the massive cybercrime problem that still plagues every major company (using AI for reviewing security control policies, account policies, and more CYBR IPO does this)
-Also related in security Intrusion detection systems have risen over 30% in sales lately, and despite defense being a lot better upgraded recently there are still many evident attack vectors. A key factor in this is companies by policy are almost always forced to use older technology, people enjoy renting out windows computers and windows computers have better virtualization server balancing support because they are the most tested)
-data content management systems that make data more available and easily constructable/displayable
-I like how you mentioned "wearables" as a seperate category. There are many things that could be wearables that are not watches most likely that people could use. A swiss army knife type thing (something that can solve many problems) that is electronic yet still carryable on a normal day basis or maybe job specific could take off.
-Using constraint provers to reason around problems in the human domain (math etc) or in computer optimization/security
-Generated business reports using machine learning + formatting (abuse free services like ipython, ihaskell, etc for output)
-Tools that keep people productive (extremely wide category there is lots of old software in it, but mostly is just old, any new way with better UI and more useful features can surpass tons of old software.)
-A "sensor world." Sensors cost near to nothing now (just a few cents under a dollar) and they can be placed basically everywhere. There needs to be a better hive-mind like use for them. Also pretty soon phone apps or simple vision type apps could be able to quickly identify objects around them using web-apps and machine learning that could provide specs or information on them as well as interact with them.
-Just simply "tech of tommorrow." Things that "can scale" massively or improve performance tech wise with speciality in that domain, though this isn't likely a first startup idea.
Note: I can list more if people want or I could expand with more specific examples of software or companies that do things in these fields etc. Feel free to reply.
There was a talk about that a while back. And the username can then be used for a bruteforce attack. News companies need to stop being so dumb and reporting on these fake hacks.
AES and RSA research has been around for ~400 years. Eliptic curves have been around for ~30 years. AES has had a huge budget for being broken and seeing as it is the predecessor to md5 which was broken before-hand it is not out of the reach of someone who can write up an implementation to use for attacks. The blackhat talk suggests that eliptic curves may be less succeptable to breach even, though it is still up to debate in the crypto communities.
There are several open source solutions to factoring and although 40 hours may seem like a long time to decrypt one single encrypted session, it will be easier with more parallel-able GPU or FPGA hardware/newer operating system abstractions that are able to compute more.
There is also more research going into making SAT solvers that are parallel so if there is a factor-able solution it may be more do-able.
Or if there is a newer breakthrough attack that comes up which does sometimes happen. Many past standards have been broken. In crypto things being broken are usually a matter of research, time, and a budget helps.
For example a linux distribution that will also make use of the GPU is currently in work: https://github.com/wbsun/kgpu
Regardless attacks will still continue to be mostly "side channel" attacks as man in the middle type attacks are usually the easiest way to subvert data exchanges.
How does this compare? What makes it specially useful for reasoning newer ideas. I've read some of the introduction about it using the PAC model, but can you give an example of how it goes about using that?
I like the idea of the language, I think it follows with Global Workspace Theory: https://en.wikipedia.org/wiki/Global_Workspace_Theory and Global Neuronal Workspace: https://en.wikipedia.org/wiki/Dehaene%E2%80%93Changeux_model
Say you compile 2 machine learning algorithm implementations. If the reverse compiler gives you an interface with a choice between say those 2 algorithms and a % accuracy rate with a visual representation this definitely sounds like a working idea of a global workspace theory. As it abstracts the complicated parts into simplistic choices that are selectable to be do-able in linear time with some statistics.
If you overlay the solved messages on the same bent surface with that type of curve you also get different results.
The reason for it being curved that way is if you stare at it from certain angles it is probably several more puzzles (based on field of view). The fourth box is probably made of more than just the 3 earlier puzzles solved but instead by different angle view puzzles. This would of course make it misleading to try to use the first three puzzles to solve the fourth.
As for content. You can definitely improve it if you spend more time on it.
The people who get hired for having good blogs have one of two things(sometimes both): good design or good content
Since your blog appears to have neither, I'd side with just listing your github, but not your blog unless you fix more of it.
As for learning AI it is best to learn it in the language that you are strongest in or a language that is easy to understand the semantics from it. For that matter Lua, Ruby, and Python are fairly commonplace for people trying to learn how to use AI. But AI algorithms can be written in any language really.
I'd suggest books with common examples of AI; like https://github.com/jbrownlee/CleverAlgorithms. He uses ruby for most of their examples but they are fairly understandable.
If you cannot get the books you want. You may be better off checking out the source codes at least. There are tons of free online books if you do some more searching. And you can even read some online ipython notebooks or blog posts about AI that may be more helpful.
Anyways, the more the merrier for discussions. So feel free to join our group. Just give me a message and I'll add you :D.
For example a hacker could introduce a security vulnerabiity in the code and then exploit it with another tool, therefore changing the original code to a certain extent.
This could also be done as a seperate website layout:
I think it'd be great to rank twitter users based on criteria: How influential are they? What category best fits them in one word? And statistics: What % of tweets have text, links, images? Filtering by media type.
I'd also see this as a very useful tool to pay for. If nothing else by ad revenue on the website itself.
Edit: Actually I think http://www.twitonomy.com/ or http://bottlenose.com/ are pretty nice.
This could be done by running an automated script that makes an e-mail address temporarily adds it to your site. Then another script that check's said e-mail address.
This is a problem with e-mail spam, but I think hiding your real e-mail from spambots is the main concern in this if I am understanding you right.
While this is true in some retrospects, it's all about solving problems. If people say "they want to solve x need for their y idea" it is essentially the same as "I'd pay for x so I can solve y sooner."
And in that retrospect these lists could be useful. You could try being less negative and think for 5-10 minutes on an idea rather then being a pessimist.