Algorithmia: App Store for algorithms

Algorithms dominate the world.

Well, the minimum algorithm rules the Internet. But since the internet is about to dominate the world, that is true. The algorithm determines that you use Google search results, the algorithm determines the Sina Weibo sidebar to show you the topic, the algorithm determines the Netflix to recommend the movie, the algorithm determines your QQ dialogue window pop-up banner ads. Technology giants spend a lot of time researching new algorithms and tweaking old algorithms. But there are some out-of-the-box algorithms that few people know about--those college computer science researchers spend years developing algorithms that are very useful to engineers, and few people outside the academic circle can see.

Diego Oppenheimer, who helped develop data analysis capabilities for tools such as Excel, Power Pivot, is too clear. He has been looking for a better algorithm to integrate into the application, but finally the public find her 1100 degrees, the person is in the dim lights-Often the answer is in Microsoft study there. So he regrets that many of the algorithms that others have studied for many years have never heard of them.

On the other hand, Oppenheimer, a university friend at USC, who is pursuing a PhD in artificial intelligence, Kenny Daniel, has another reason: Although he has published several popular academic algorithms, these algorithms do not have a chance to apply to the actual. Therefore, the two hit, decided to partner to solve the common problems between the two sides. Build an algorithm for the "App Store"-algorithmia.

The idea is to give algorithmic inventors the opportunity to apply their work to the real world and get money, while also making it easier for other companies without Microsoft and Google to get into the algorithm and find the best solution for their problems.

Currently, there have been some algorithmic markets, such as the Dataxu of advertising algorithms, selling the snapanalytics of prefabricated predictive models and Algorithms.io focusing on machine learning algorithms. But Algorithmia's difference is that it accepts and sells all types of algorithms. The other is Stephen Wolfram's famous Mathematica Online, the cloud service's math software integrates many algorithms, but it's more of a home-grown algorithm.

Mechanism

Can the algorithm sell? How does this thing avoid being copied?

Building an algorithmic market is not easy. In fact, even if you can find an algorithm to use in an academic paper, the implementation of the algorithm is often difficult. The authorization algorithm is also a problem for users.

Algorithmia to solve these two problems by cloud service. Once the user found the algorithm to use, just add a few lines of simple algorithm query code to the application, Algorithmia Server will guide the application query to the algorithm implementation, return the result, thus avoiding the need to integrate the algorithm into application. The cost of the query depends on the algorithm author's pricing.

In addition, the algorithm authors can choose to open their own algorithms, so that developers can directly integrate the algorithm into their applications, but algorithmia hope that the convenience of cloud services to make it worthwhile users pay. However, the disadvantage is that the user must send the data to the Algorithmia server for processing. As a result, there is a risk of data exposure, in addition to the large amount of data may have a performance problem.

Algorithmia's approach is to make resources available to a number of cloud service providers, such as Amazon, Rackspace, and so on, to ensure the performance of most users. In addition, they plan to offer local editions to companies with data centers.

The algorithm of Yelp

Analyst and trainer Ajay Ohri look forward to similar things for years. Now many mobile app developers can make a lot of money from app stores, but although algorithmic development is difficult and useful, algorithmic developers are hard to make money. So he always wanted a place where people could make money, even if only a little bit of application developers.

But given that Algorithmia is still in the private testing phase, whether the service is the solution he is looking for is unknown. His hope is that algorithmia can develop the imaging reviews site Yelp as the algorithm of the comments site. The idea is to help developers choose the best algorithm even more important than the market component itself, because the most suitable algorithm is very difficult to determine. And Algorithmia does have a rating and comment function for a similar app store.