Google has tweaked its "deep learning" AI to use an external memory bank. It's an attempt to replicate the way human brains use short term memory to simplify reasoning. The company demonstrated the approach by having the system teach itself the London Underground (subway) map and figure out the quickest route between stops. It's a simple task to humans, but the process – which involves comparing multiple branching options with 270 stops over 11 lines – is exactly the type of problem that poses a challenge to artificial intelligence. Because the system was allowed temporary access to stored memory, it was able to more effectively process and categorize the possible routes without having to start from scratch each time. That's similar to how a human brain could use short term memory to filter down all the possible routes by ruling out every one that involves travelling in a particular direction from a specified stop, repeating the process until only the optimum answer remained.

We think the next era will belong to "the conversational layer" -- both text- and voice-driven -- that will use chat, messaging, or natural language interfaces to interact with people, brands, services, and bots. This interaction will occur at the exact time the user demands a product or service, and in the exact terms she thinks of that product or service, in the language and communication methods she typically uses (intent, words, shortcuts, emojis, etc.). In order to reach user conversations today, brands will need to decide which platforms to target and build on. Many of the tools that are provided by the messaging and bot platform providers are from the open-source space, and companies can perform low-cost experiments with a reduced set of users to learn more about conversational interactions and use cases that yield the desired results.

Originally posted here, where you can see all the graphics. There has been much in the news lately about the next wave of MT technology driven by a technology called deep learning and neural nets (DNN). I will attempt to provide a brief layman's overview about what this is, even though I am barely qualified to do this (but if Trump can run for POTUS then surely my trying to do this is less of a stretch). Please feel free to correct me if I have inadvertently made errors here. Traditional AI methods of language understanding depended on embedding rules of language into a system, but in the Google SmartReply project, as with all modern machine learning, the system was fed enough data to learn on its own, just as a child would.

Backpropagation is a common method for training a neural network. There is no shortage of papers online that attempt to explain how backpropagation works, but few that include an example with actual numbers. This post is my attempt to explain how it works with a concrete example that folks can compare their own calculations to in order to ensure they understand backpropagation correctly. If this kind of thing interests you, you should sign up for my newsletter where I post about AI-related projects that I'm working on. You can play around with a Python script that I wrote that implements the backpropagation algorithm in this Github repo.

The goal of predictive modeling is to create models that make good predictions on new data. We don't have access to this new data at the time of training, so we must use statistical methods to estimate the performance of a model on new data. This class of methods are called resampling methods, as they resampling your available training data. In this tutorial, you will discover how to implement resampling methods from scratch in Python. How to Implement Resampling Methods From Scratch In Python Photo by Andrew Lynch, some rights reserved.

I've worked on Data Science projects and delivered Machine Learning models both in production code and more research type work at a few companies now. Some of these companies were around the Seed stage/ Series A stage and some are established companies listed on stock exchanges. The aim of this article is to simply share what I've learned -- I don't think I know everything. I think my audience consists of both managers and technical specialists who've just started working in the corporate world -- perhaps after some years in Academia or in a Startup. My aim is to simply articulate some of the problems, and propose some solutions -- and highlight the importance of culture in enabling data science.

Today, I am starting a new series! You may notice that I recently finished a Machine Learning course on Coursera (definitely recommend it). But here comes the real challenge, put my skill into real use! I'm going to participate in the local project competition so-called SO? (i.e. So, I'm starting a new series about my project related to the machine learning.

Talend is the only data integration platform that supports the latest Hadoop Distribution. Native Spark connectors in Talend optimize data feeds from external sources into Spark so you can ingest, load in parallel, and accelerate use of data.

In engineering, there are various ways to build a key-value storage, and each design makes a different set of assumptions about the usage pattern. In statistical modeling, there are various algorithms to build a classifier, and each algorithm makes a different set of assumptions about the data. When dealing with small amounts of data, it's reasonable to try as many algorithms as possible and to pick the best one since the cost of experimentation is low. But as we hit "big data", it pays off to analyze the data upfront and then design the modeling pipeline (pre-processing, modeling, optimization algorithm, evaluation, productionization) accordingly. As pointed out in my previous post, there are dozens of ways to solve a given modeling problem. Each model assumes something different, and it's not obvious how to navigate and identify which assumptions are reasonable.