IT analysis: Forecasts for 2019

Chema Dec. 21, 2018

security 2019 forecast machine learning blockchain privacy

In this year that has ended, the protagonists have been the increase of the machine learning, the fall of the cryptocurrencies and the scandals for lack and failures in the privacy. 2019? These are our forecasts.

Machine Learning for All

The democratization of machine learning will begin being a reality in 2019. 

As you know, the challenge today is in the preparation and adaptation of data to train neural networks.

The development of the GPUs, driven mainly by the exponential growth that it supposed the mining of cryptocurrencies like Ethereum in 2017.

In addition, the fall of cryptocurrencies in 2018 has led to helping nVidia fall in the stock market, therefore there is an offer in second-hand market of GPUs with large capacity for processing and memory.

A few years ago, Google helped democratize the state-of-the-art of machine learning and deep learning, with the release of the framework, TensorFlow.

This allows anyone to train an image classification system at home. In fact, you could even download the trained network model and therefore saving time and watts.

Even so, the tensorflow learning curve has a medium complication. For that reason, frameworks such as Keras were born, which uses TensorFlow as a backend and allows programming from Python models in a few lines of code (very few lines of code).

Although you can simplify the thing more, sometimes it is not easy to find the network model and which parameters are the most appropriate.

But this problem has been solved with AutoML, developed by Google, which allows you to automatically find the best model for your machine learning problem. Then they created AutoKeras, which offers the same but based on Keras. With AutoKeras now a problem of machine learning is solved with little more than 3 lines of code.

	
	import autokeras as ak

clf = ak.ImageClassifier()
clf.fit(x_train, y_train)
results = clf.predict(x_test)

Following this line of evolution, it is not surprising that during 2019 we see how services and start-ups are born that offer "create your own machine learning" services in 3-clicks. We will see some new acronyms: MLaaS. It will happen? We will see it.

Privacy-first Services

In the middle of 2018 the presence of a guest actor was confirmed, privacy, or it would be better to say the lack of it.

Scandals like Facebook's with Cambridge Analytica, the erroneous sending of voice transcripts exchanged with Alexa to another user or the hacking of passwords like the Quora question and answer service, expose security and privacy errors.

More and more users of Facebook and other "free" Internet services, realize one thing: they are not free, their data are worth money and there are even tools that allow estimating their value.

Many users have started to leave social networks and they have to start worrying more about security and privacy.

A while ago we started talking about the "Enterprise Ready". It consists in bringing apps that are usually used in the cloud to the servers of the organization or company.

We believe that the trend for this 2019 will be to use cloud which is a mix of public (AWS, Google Cloud, Azure, ...) and private cloud services. DevOps technology has evolved a lot, and services such as Kubernetes or Ansible will allow to treat this cloud as a single one and in a transparent way.

2018 ended with Amazon making an expensive bet in Spain to sell and position systems with Alexa. This strategic movement consists from the sale of their own devices such as Echo Dot, to the integration of third parties such as the Sonos speakers, and of course, the promotion of competitions where they liven developers up to create apps (content) for Alexa.

The point is that there are already many voices that warn of the danger of having amazon "always" listening. If you already have a highly developed client profile, imagine when you ask everything.

So far this year, Apple already boasts privacy and has done this campaign in full CES 2019, "what happens on your iPhone, stays on your iPhone".

Apple's ad on wall near of CES 2019, Las Vegas

For this reason, we believe that services like Mycroft are going to be successful. Its value proposition is to have an Alexa but without Amazon and of course, open source. This means that anyone can install Mycroft at home and the data will stay at home (yes, you do not have to worry about the security of your data).

In short, we believe that those services that take into account privacy and security in their value proposition, will have a great advantage over competitors that does not have it.

Blockchain: the killer App

It is said that WhatsApp was the definitive app that boosted the sales of the smartphone (especially Android). It is what is known as a killer app. People without any contact with technology go to mobile stores looking for "a mobile with WhatsApp".

In the world of blockchain, it is clear that it has not happened. There is not such app yet.
The closest thing we known was CryptoKitties, which was dedicated to raise and sell under the network Ethereum unique digital cats (¬¬).

In 2018 it became clear that the gold rush for cryptocurrencies passed.
In our opinion, that investment whose value is not understood and only used for speculating does not go very far. Even the fever for tulips made more sense: a tulip can be touched, you can see it and you can smell it.

Now that the speculators are out, for the time being, the first blockchain or DLT apps start to appear.

It is already used in environments such as the University to carry out a record of authorship of publications. It is also used in governance environments and banks have begun to record transactions in private networks with them.

We do not know if 2019 will be the year where that killer app for blockchain appears and consolidates, but it will be the year in which the candidate apps appear to be. We will be alert.

Our bet: Machine Learning applied to your needs

For some time now we have been testing and developing technology/machine learning algorithms.

I think we are in the position of applying that know-how in our clients' projects.

Therefore, if you need to automate some task of your business, let us evaluate if a machine-based solution adapted to your needs would be feasible. We will do it willingly.

We hope that this 2019 will be profitable and all your hopes will be digitalized. Count on us!

(Este post está disponible en castellano en Medium)