September 25, 2016

By starting with a contract or test case that is well understood to incrementally test a feature requirement, you ensure that as a small iterative unit of work completes, it meets that contract in such a way that it is releasable, deployable software. Exploratory testing tools for new feature development come to play as do coverage tools that send data showing anomalies between releases back to the quality process. Coveralls.io is a great tool that’s easy to configure and has wonderful visualizations for the most popular languages, while Jenkins has a highly customizable dashboard. ... Technology can’t solve all problems, however, so developers and testers will need to change some of their workflows to master CT. These concepts are closely linked to the Agile and DevOps practices you are probably already using, so adapting testing in this way should not be a huge shift.


Google Allo: Don't use it, says Edward Snowden

Allo does support end-to-end encryption, which should make it difficult for anyone but recipient and sender to view the contents of messages; however, Google was criticized by Snowden and other privacy advocates for setting it as off by default. Allo relies on the encryption protocol used by Signal, which Snowden has vouched for as a private messaging app, but in Allo it is only active when users are in Incognito Mode. "We've given users transparency and control over their data in Google Allo. And our approach is simple -- your chat history is saved for you until you choose to delete it. You can delete single messages or entire conversations in Allo," Google said in a statement toTechCrunch.


Investing in AI offers more rewards than risks

While some may argue it’s impossible to predict whether the risks of AI applications to business are greater than the rewards (or vice versa), analysts predict that by 2020, 5 percent of all economic transactions will be handled by autonomous software agents. The future of AI depends on companies willing to take the plunge and invest, no matter the challenge, to research the technology and fund its continued development. Some are even doing it by accident, like the company that paid a programmer more than half a million dollars over six years, only to learn he automated his own job. Many of the AI advancements are coming from the military. The U.S. government alone has requested $4.6 billion in drone funding for next year, as automated drones are set to replace the current manned drones used in the field.


Crowdsourcing Data Governance

This variation of context is why the right operating model set up is so important for any data governance initiative, especially the ones that are just getting started. A successful data governance initiative will bring change, and so time becomes yet another dimension for the context. I’ve seen it happen many times: organizations launch with a best-in-class operating model to drive their stewardship. They gain adoption, and the resulting change makes the original operating model obsolete, or rather stretches it to the limit. This is why I am absolutely convinced that a data governance platform that aims to be successful needs a capability for operating model configuration: your roles, responsibilities, workflows, dashboards, views, use cases, and more.


Bossie Awards 2016: The best open source application development tools

or years and years, we’ve been building applications that collect data from the users and serve it back to them. We’re finally starting to do something with that data. Along with the best open source tools for building web apps, native apps, native mobile apps, and robotics and IoT apps, this year’s Bossie winners in application development include top projects for data analysis, statistical computing, machine learning, and deep learning. After all, if our applications can be reactive, responsive, and even “ambitious,” they can also be intelligent.


Is this the age of Big OLAP?

What has dogged OLAP, though, is its scalability. Most OLAP servers run on single, albeit beefy, servers, which limits the parallelism that can be achieved and therefore imposes de facto limits on data volumes. Customers who hit these scalability ceilings may contemplate using Big Data technologies, like Hadoop and Spark, but those tend not to employ the dimensional paradigm to which OLAP users are accustomed. What to do? Well, a few vendors have decided to take Hadoop and Spark, and leverage them as platforms on which big OLAP cubes can be run and built. ... Their approach has been to let people in those enterprises work in the OLAP environments they are comfortable with and, at the same time, make use of their Hadoop clusters.


Deep Learning in a Nutshell: Reinforcement Learning

Reinforcement learning is about positive and negative rewards (punishment or pain) and learning to choose the actions which yield the best cumulative reward. To find these actions, it’s useful to first think about the most valuable states in our current environment. For example, on a racetrack the finish line is the most valuable, that is the state which is most rewarding, and the states which are on the racetrack are more valuable than states that are off-track. Once we have determined which states are valuable we can assign “rewards” to various states. For example, negative rewards for all states where the car’s position is off-track; a positive reward for completing a lap; a positive reward when the car beats its current best lap time; and so on.


NYDFS Proposed Cybersecurity Regulation for Financial Services Companies

The goal of the Proposed Regulation is to secure “Nonpublic Information” from misuse, disruption and unauthorized access, and as noted above, such information is defined broadly. It includes not only competitively sensitive information and intellectual property, but also numerous categories of information that a Covered Entity receives from or about consumers, including information considered nonpublic personal information under the GLBA Privacy Rule. ...” When something goes wrong, the Covered Entity must report it to the Superintendent. Specifically, any attempt or attack “that has a reasonable likelihood of materially affecting the normal operation of the Covered Entity or that affects Nonpublic Information” must be reported to the Superintendent within 72 hours after the Covered Entity becomes aware of the event.


Big Data Processing with Apache Spark - Part 5: Spark ML Data Pipelines

Machine learning pipelines are used for the creation, tuning, and inspection of machine learning workflow programs. ML pipelines help us focus more on the big data requirements and machine learning tasks in our projects instead of spending time and effort on the infrastructure and distributed computing areas. They also help us with the exploratory stages of machine learning problems where we need to develop iterations of features and model combinations. Machine Learning (ML) workflows often involve a sequence of processing and learning stages. Machine learning data pipeline is specified as a sequence of stages where each stage is either a Transformer or an Estimator component. These stages are executed in order, and the input data is transformed as it passes through each stage in the pipeline.


How fintech startups can disrupt the financial services industry

Successful fintech startups will embrace “co-opetition” and find ways to engage with the existing ecosystem of established players. E.g. PayPal partners with Wells Fargo for merchant acquisition. Some business lending platforms enable banks to participate as credit providers on their platforms. Conversely, some banks partner with P2P lending platforms to provide credit to those borrowers who would ... Fintech startups are flying under the regulatory radar so far. However that may change in the near future. Regulatory tolerance for lapses on issues such as know your customer, compliance, and credit-related disparate impact will be low. Experience of microfinance industry in many developing countries the past is a good indicator of the high impact of regulation on an unregulated industry.



Quote for the day:


"It is better to be defeated standing for a high principle than to run by committing subterfuge." -- Grover Cleveland