We have just released version 4.2 of the Jupyter notebook package. There are the usual bugfixes, and a few improvements. You can see a summary of changes in the docs, or see all the changes on GitHub.

Altogether, 21 authors contributed to 53 pull requests to make this release.

The primary feature of 4.2 is improvements to the extension system for the notebook. The notebook can be extended in two ways:

1. nbextensions, which are JavaScript extensions that run in your browser. These can modify the UI or behavior of the notebook, such as adding buttons to your toolbar, new keyboard shortcuts, etc.
2. serverextensions, which modify the Python webserver. These can add new HTTP endpoints, register extra events to run when you save your notebook, etc.

The improvements are focused on making installing and enabling extensions easier, including:

1. better symmetry between handling of serverextensions and nbextensions
2. allowing enabling nbextensions system-wide
3. working better with the Python sys.prefix for installing/enabling extensions using packages and environments, such as conda packages or python wheels
4. allowing installation via Python packages

Source: https://blog.jupyter.org/2016/04/15/notebook-4-2/

Tinker with a real neural network right here in your browser.

Um, What Is a Neural Network?

It’s a technique for building a computer program that learns from data. It is based very loosely on how we think the human brain works. First, a collection of software “neurons” are created and connected together, allowing them to send messages to each other. Next, the network is asked to solve a problem, which it attempts to do over and over, each time strengthening the connections that lead to success and diminishing those that lead to failure. For a more detailed introduction to neural networks, Michael Nielsen’s Neural Networks and Deep Learning is a good place to start. For more a more technical overview, try Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron Courville.

Source: A Neural Network Playground

Keras was initially released a year ago, late March 2015. It has made tremendous progress since, both on the development front, and as a community.

But continuous improvement isn’t enough. A year of developing Keras, using Keras, and getting feedback from thousands of users has taught us a lot. To the point that we are now able to redesign it better than we could have the first time around.

And so today we are releasing Keras 1.0. It isn’t a patch on top of the previous version, it is actually a re-writing of Keras nearly from scratch. It maintains backwards compatibility while introducing a series of major features, made possible by a better design under the hood.

Simplicity and accessibility have always been the targets guiding the Keras development efforts. The purpose of Keras is to make deep learning accessible to as many people as possible, by providing a set of “Lego blocks” for building Deep Learning models in a fast and simple way. Keras 1.0 pushes even further in that same direction.

The most significant feature introduced today is the functional API, a new way to define your Keras models. Get started with the functional API with this short guide. If you are new to Keras, first read the “30 seconds to Keras” introduction, then read this overview of the Sequential model.

New features

• The functional API: a simpler and more powerful way to define complex deep learning models.
• Better performance. Compilation times are lower. All RNNs now come in 2 different implementations to choose from, allowing you to get maximum performance across widely different tasks and setups. And Theano RNNs can now be unrolled, yielding up to a 25% speed-up.
• Modular metrics. You can know monitor arbitrary lists of metrics on arbitrary endpoints of your Keras models.
• An even better user experience. The code has been rewritten from scratch with the end user in mind at all stages. A great library UX has two components: simple, intuitive APIs (the kind that are easy to memorize), and the ability to return sensible, easy to grok error messages whenever faced with a user error.
• Improved Lambda layers.

…and much more.

Source: Introducing Keras 1.0

An analysis of a Neanderthal Y chromosome suggests human hybrids containing it would have been unviable, and explain why it is not found in modern humans

Modern humans diverged from Neanderthals some 600,000 years ago – and a new study shows the Y chromosome might be what kept the two species separate.

It seems we were genetically incompatible with our ancient relatives – and male fetuses conceived through sex with Neanderthal males would have miscarried. We knew that some cross-breeding between us and Neanderthals happened more recently – around 100,000 to 60,000 years ago.

Neanderthal genes have been found in our genomes, on X chromosomes, and have been linked to traits such asskin colour, fertility and even depression and addiction. Now, an analysis of a Y chromosome from a 49,000-year-old male Neanderthal found in El Sidrón, Spain, suggests the chromosome has gone extinct seemingly without leaving any trace in modern humans.

This could simply be because it drifted out of the human gene pool or, as the new study suggests, it could be because genetic differences meant that hybrid offspring who had this chromosome were infertile – a genetic dead end.
Source: Missing Y chromosome kept us apart from Neanderthals | New Scientist

You’ve seen The Wolf of Wall Street, you’re handy with some code — you just know you can beat the market, but you don’t have the money to invest. What could possibly go wrong. Quantiacs has the platform for you to experiment and try out strategies — and the money to back your bets, too.

At the moment, Quantiacs is dogfooding its own systems; the founders have $3.75 million of their own money invested in the platform, running across six different algorithms. Quantiacs has enough faith in its system that it’s publishing the live performance of the algorithms on its website, and the team says it expects to open the platform to external investors starting in May this year.

The easiest way to explain what Quantiacs does is to think of it as crowdsourcing for trading algorithms. The company empowers “quants” — people who create quantitative algorithms — to participate in the industry by connecting their algorithms to capital from institutional investors.

Source: Quantiacs is high-stakes fantasy football for quants | TechCrunch

Software development begins as a quest for capability, doing what could not be done before. Once that what is achieved, the engineer is left with thehow. In enterprise software, the most frequently asked questions are, “How fast?” and more importantly, “How reliable?”

Questions about software performance cannot be answered, or even appropriately articulated, without statistics.

Yet most developers can’t tell you much about statistics. Much like math, statistics simply don’t come up for typical projects. Between coding the new and maintaining the old, who has the time?

Engineers must make the time. I understand fifteen minutes can seem like a big commitment these days, so maybe bookmark it. Insistent TLDR seekers can head for our instrumentation section or straight to the summary.

For the dedicated few, class is in session. It’s time to broach the topic, learn what works, and take the guesswork out of software. A few core practices go a long way in generating meaningful systems analysis. And a few common mistakes set projects way back. This guide aims to lighten software maintenance and speed up future development through answers made possible by the right kinds of applied statistics.

Source: Statistics for Software | PayPal Engineering Blog

Data science is the sexiest job of the 21st century. Data scientists around the world are presented with exciting problems to solve. Within the complex questions they have to ask, a growing mountain of data rests a set of insights that can change entire industries. In order to get there, data scientists often rely on programming languages and tools.

This is an excerpt of our free, comprehensive guide to getting a job in data science that deals with two of the most common tools in data science, Python and R.

Python

Python is a versatile programming language that can do everything from data mining to plotting graphs. Its design philosophy is based on the importance of readability and simplicity. From the The Zen of Python:

• Beautiful is better than ugly.
• Explicit is better than implicit.
• Simple is better than complex.
• Complex is better than complicated.
• Flat is better than nested.
• Sparse is better than dense.
• Readability counts.

As you can imagine, algorithms in Python are designed to be easy to read and write. Blocks of Python code are separated by indentations. Within each block, you’ll discover a syntax that wouldn’t be out of place in a technical handbook.

Source: Data science sexiness: Your guide to Python and R, and which one is best

Discover fresh relevant content to your interests, save interesting articles, follow influential experts, be the first to share soon-to-be viral content and much more

Source: bioinformatics Articles : GWASdb2 – A database for human genetic variants identified by genome-wide association studies | bioinformatics-databases

The other day I saw a tweet about blog posts it would be great to see, and one of them was “inside PostgreSQL: how a join works internally, what makes it fast”. Coincidentally, I’m taking a course on database system implementation, and I thought this would be a great way to study for my midterm! Of course, by the time I’ve finished this post, the midterm is long gone, but it was still a great incentive for me to start writing 🙂

Source: PostgreSQL Query Optimization · Jincheng Li