Mitigating offensive search suggestions with deep learning 06-12

Image credit : Shyam's Imagination Library

The humble search bar is the window through which most Internet users experience the web. Deep
learning is set to enhance the capabilities of this simple tool such that search engines can now anticipate what the user is looking for whilst moderating offensive suggestions before the query is complete.

A lack of contextual and deeper understanding of the intent of search queries often leads to inappropriate or offensive suggestions. Striving for a safer and saner web, the Microsoft team began its research with deep learning techniques to help detect and automatically prune search suggestions.

Our paper titled “Convolutional Bi-Directional LSTM for Detecting Inappropriate Query Suggestions in Web Search” received the “Best Paper Award” at the recent Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) 2017. It was picked amongst a record-breaking 458 submissions at the leading international conference on knowledge discovery and data mining. Winning the award has been a humbling experience for the team. It upholds our efforts to establish deep machine learning as a powerful weapon against online vitriol.

Here’s a brief overview of what we found:

The Challenge

Human beings intuitively detect offensive language and derogatory speech online. Algorithms, however, struggle to clearly distinguish toxic language from benign comments. Systems often misjudge the level of toxicity or hate in certain phrases that require a deeper understanding of cultural norms, slang and context.

The problem is usually rendered difficult due to unique challenges posed by search queries such as lack of sufficient context, natural language ambiguity and presence of spelling mistakes and variations. For example, Marvin Gaye’s classic hit ‘If I Should Die Tonight’ could be deemed offensive simply because of the inclusion of the phrase ‘should die tonight’. Similarly, algorithms frequently misclassify offensive words and phrases as ‘clean’ simply because they were misspelled or used euphemisms. For example, the phrase ‘shake and bake’ is both a registered trademark for a popular food brand and street code for preparing illegal drugs on the move.

Safeguarding the Online Experience

The impact of inappropriate language and offensive speech online cannot be overstated. Internet access is ubiquitous and users cover diverse age groups and cultural backgrounds. Making search queries, online communication, and instant messaging safe for children, minorities, and sensitive communities is essential to preserve the integrity of the digital world.

Inappropriate suggestions on search queries or offensive comments on news articles could cause significant harm to vulnerable groups such as children and marginalized communities. Unsuitable suggestions could tarnish the reputation of corporations, inadvertently help someone cause harm to themselves with risky information, or lead to legal complications with authorities and regulators. Problems such as intimidation, threats, cyber bullying, trolling, explicit and suggestive content, and racist overtones need to be curtailed to help keep the Internet open and safely accessible to everyone.

The Solution

Conventional solutions to this problem have typically involved using –

1. A manually curated list of patterns involving such offensive words, phrases and slangs or
2. Classical Machine Learning (ML) techniques which use various hand-crafted features (typically words etc.) for learning the intent classifier or
3. Standard off-the-shelf deep learning model architectures such as CNN, LSTMs or Bi-directional LSTMs (BLSTMs).

In our current work, we propose a novel deep learning architecture called, “Convolutional Bi-Directional LSTM (C-BiLSTM)” – which combines the strengths of Convolutional Neural Networks (CNNs) with Bi-directional LSTMs (BLSTMs). Given a query, C-BiLSTM uses a convolutional layer for extracting feature representations for each query word which is then fed as input to the BLSTM layer which captures the various sequential patterns in the entire query and outputs a richer representation encoding them. The query representation thus learnt passes through a deep, fully connected network which predicts the target class – whether it is offensive or clean. C-BiLSTM doesn’t rely on hand-crafted features, is trained end-to-end as a single model, and effectively captures both local features as well as their global semantics.

Applying the technique on 79041 unique real-world search queries along with their class labels (inappropriate/clean), revealed that this novel approach was significantly more effective than conventional models based on patterns, classical ML techniques using hand-crafted features. C-BiLSTM also outperformed standard off-the-shelf deep learning models such as CNN, LSTM and BLSTM when applied to the same dataset. Our final C-BiLSTM model achieves a precision of 0.9246, recall of 0.8251 and an overall F1 score of 0.8720.

Although the focus of the paper was detecting offensive terms in Query Auto Completion (QAC) in search engines, the technique can be applied to other online platforms as well. Comments on news articles can be cleaned up and inappropriate conversations can be flagged up for abuse. Trolling can be detected and a safe search experience can be enabled for children online.  This technique could also help make chatbots and autonomous virtual assistants more contextually aware, culturally sensitive, and dignified in their responses. More details about this technique and implementation could be found in the actual paper.

Final Thoughts

The deep learning technique detailed in this study could be a precursor to better tools that can fight online vitriol. When APIs based on this system are applied to social media platforms, email services, chat rooms, discussion forums, and search engines, the results will be parsed through multiple filters to ensure users are not exposed to offensive content.

Curtailing offensive language can transform the Internet, making it safely accessible to a wider audience. Making the web an open, safe and secure place for ideas and innovations is a core part of our mission to empower everyone, everywhere through the power of technology.

Tags best paper award C-BiLSTM deep learning machine learning PAKDD 2017 award query auto completion (QAC)  

View at the original source

ARM’s new processors are designed to power the machine-learning machines 06-06

ML plus AI, AR, and VR — ARM is taking on the full set of trendy initialisms

On the eve of Computex, Taiwan’s big showpiece event where PC makers roll out the latest and best implementations of Intel CPUs, mobile rival ARM is announcing its own big news with the unveiling of a new generation of ARM CPUs and GPUs. Official today, the ARM Cortex-A75 is the new flagship-tier mobile processor design, with a claimed 22 percent improvement in performance over the incumbent A73. It’s joined by the new Cortex-A55, which has the highest power efficiency of any mid-range CPU ARM’s ever designed, and the Mali-G72 graphics processor, which also comes with a 25 percent improvement in efficiency relative to its predecessor G71.

The efficiency improvements are evolutionary and predictable, but the revolutionary aspects of this new lineup relate to artificial intelligence: this is the first set of processing components designed specifically to tackle the challenges of onboard AI and machine learning. Plus, last year’s updates to improve performance in the power-hugry tasks of augmented and virtual reality are being extended and elaborated.

Before we dive into the detail of this year’s changes, it’s worth recapping what ARM does and why it’s important. This English company, now owned by Japan’s SoftBank, is responsible for designing the processor architecture of practically every mobile device — you’ll have heard of Qualcomm’s Snapdragon, Samsung’s Exynos, and Apple’s A-series of mobile chips, all of which are built using ARM’s instruction sets and based on ARM’s design blueprints. When we talk about the oncoming wave of mobile AI, mobile VR, and smartphones that can perform machine-learning tasks without sending them off to processor farms up in the cloud, developing the capabilities for those tasks starts with ARM. 

The new Cortex-A75 and A55 are the first Dynamiq CPUs from ARM. Dynamiq is the branding chosen to describe a much more flexible set of design options for silicon vendors like Qualcomm. Where previously ARM allowed for designs that paired a cluster of so-called big CPUs (from its A7x class) and a matched number of little CPUs (from the A5x series), the new design makes it possible to spec a single, mixed-up cluster composed of both big and little CPUs, to a maximum of eight. Thus, chip makers can now have, for example, seven little A55 cores and just one big A75 one: for a favorable mix of long battery life, cost efficiency, and a high ceiling of single-threaded performance when it’s called for.

"50x improvement in AI performance over the next three to five years"

A RM marketing chief John Ronco says he anticipates a "50x improvement in AI performance over the next three to five years thanks to better architecture, micro-architecture, and software optimizations." ARM’s Dynamiq changes include a redesigned memory subsystem and tweaks to how CPU caches work — which has led to a doubling of memory streaming performance on the A55 relative to the A53 preceding it. Given that the A53 has shipped on 1.7 billion devices over the past three years, it’s truly the A55 that will make the biggest difference in achieving Ronco’s ambitious forecast. In most applications, the new mid-range core will be 10 to 30 percent better than previously, offering up to 15 percent better power efficiency and 18 percent better single-thread performance.

But it’s the fact that the new chip designs will be 10 times more configurable, with up to 3,000 different configurations, that will allow chipmakers far greater flexibility to make the most of them by tailoring them to specific tasks.

Interestingly, ARM won’t just be powering machine learning with its new chips, it’ll benefit from ML too. The new designs benefit from an improved branch predictor that uses neural network algorithms to improve data prefetching and overall performance.

The Cortex-A75 makes double-figure performance improvements across the board, with ARM claiming it’s on average 22 percent better than the A73, with 16 percent higher memory throughput, and a 34 percent improvement in its Geekbench score. Single-threaded performance, according to ARM’s Ronco, is up by 20 percent, purely by improving the instructions-per-clock efficiency. The A75 chip is roughly 2.5x the size of the A55, and its intended uses are for infrastructure, automotive, and rich mobile applications. Yes, that means VR, AR, and high-fidelity games, the latter of which ARM’s research has shown have been rapidly increasing in popularity.

A major architectural change with the A75 is the opening of a larger power envelope for chips using this core, scaling up to 2W of power consumption, and thus offering up to 30 percent of extra performance on larger-screen devices. This is entirely targeted at the upcoming Windows on ARM reboot, expected later this year. It’s worth noting that in ARM’s world a "large" screen basically amounts to a laptop — and the company set up a dedicated Large Screen Compute division a year and a half ago to more aggressively target the clamshell devices that Intel has been dominant in. 

As to the new Mali GPU, it has 32 shader cores, 25 percent higher energy efficiency, and a 20 percent better performance density (aka performance per mm² of space). The Mali-G72 is at the heart of ARM’s push toward improving machine learning efficiency, and ARM claims it’s showing itself to be 17 percent better than the G71 in ML benchmarks. The design optimizations from the company are tailored to accelerate inference engines rather than training engines — that’s to say ARM chips will be best at using accumulated ML capabilities rather than developing them, which makes perfect sense for mobile applications. Training AI will be a task better left to Nvidia and AMD graphics cards or Google’s custom TensorFlow TPUs.

The Cortex-A75 and A55 designs were released to ARM’s partners at the end of 2016, so by this point, they’ve all had a few months to decide what to do with them. ARM says a "realistic time window" for new mobile devices powered by its latest designs would be the first quarter of 2018 — though the company is also conscious of a new phenomenon it describes as "China speed," where Chinese phone vendors will put its designs into products almost immediately. The Huawei Mate 9, for example, was released just eight months after ARM distributed the Mali-G71 to partners. This faster Chinese cadence could lead to some A75- and A55-based designs this year, but then bulk of them are likely to arrive with the usual smartphone refresh cycle early next year.



View at the original source

The Democratization of Machine Learning: What It Means for Tech Innovation 04-15

The world of high-tech innovation can change the destiny of industries seemingly overnight. Now we are on the cusp of a new grand leap thanks to the democratization of machine learning, a form of artificial intelligence that enables computers to learn without being explicitly programmed. This process of democratization is already underway.

                                     Image credit : Shyam's Imagination Library

Last month, at the CloudNext conference in San Francisco, Google announced its acquisition of Kaggle, an online community for data scientists and machine-learning competitions. Although the move may seem far removed from Google’s core businesses, it speaks to the skyrocketing industry interest in machine learning (ML). Kaggle not only gives Google access to a talented community of data scientists, but also one of the largest repositories of datasets that will help train the next generation of machine-learning algorithms.

As ML algorithms solve bigger and more complex problems, such as language translation and image understanding, training them can require massive amounts of pre-labeled data. To increase access to such data, Google had previously released a labeled dataset created from more than 7 million YouTube videos as part of their YouTube-8M challenge on Kaggle. The acquisition of Kaggle is an interesting next step.

1. Highly scalable computing platforms
2. Even if specialized processors were available, not every company has the capital and skills needed to manage a large-scale computing platform needed to run advanced machine learning on a routine basis. This is where public cloud services such as Amazon Web Services (AWS), Google Cloud Platform, Microsoft Azure and others come in. These services offer developers a scalable infrastructure optimized for ML on rent and at a fraction of the cost of setting up on their own.
3. Open-source, deep-learning software frameworks

A major issue in the wide-scale adoption of machine learning is that there are many different software frameworks out there. Big companies are open sourcing their core ML frameworks and trying to push for some standardization. Just as the cost of developing mobile apps fell dramatically as iOS and Android emerged as the two dominant ecosystems, so too will machine learning become more accessible as tools and platforms standardize around a few frameworks. Some of the notable open source frameworks include Google’s TensorFlow, Amazon’s MXNet and Facebook’s Torch.

4. Developer-friendly tools

The final step to democratization of machine learning will be the development of simple drag-and-drop frameworks accessible to those without doctorate degrees or deep data science training. Microsoft Azure ML Studio offers access to many sophisticated ML models through a simple graphical UI. Amazon and Google have rolled out similar software on their cloud platforms as well.

5. Marketplaces for ML algorithms and datasets

Not only do we have an on-demand infrastructure needed to build and run ML algorithms, we even have marketplaces for the algorithms themselves. Need an algorithm for face recognition in images or to add color to black and white photographs? Marketplaces like Algorithmia let you download the algorithm of choice. Further, websites like Kaggle provide the massive datasets one needs to further train these algorithms.

“The final step to democratization of machine learning will be the development of simple drag-and-drop frameworks accessible.”

All of these changes mean that the world of machine learning is no longer restricted to university labs and corporate research centers that have access to massive training data and computing infrastructure.

What are the implications?

Back in the mid- and late-1990s, web development was done by specialists and was accessible only to firms with ample resources. Now, with simple tools like WordPress, Medium and Shopify, any lay person can have a presence on the web. The democratization of machine learning will have a similar impact of lowering entry barriers for individuals and startups.

Further, the emerging ecosystem, consisting of marketplaces for data, algorithms and computing infrastructure, will also make it easier for developers to pick up ML skills. The net result will be lower costs to train and hire talent. We think that the above two factors will be particularly powerful in vertical (industry-specific) use cases such as weather forecasting, healthcare/disease diagnostics, drug discovery and financial risk assessment that have been traditionally cost prohibitive.

Just like cloud computing ushered in the current explosion in startups, the ongoing build-out of machine learning platforms will likely power the next generation of consumer and business tools. The PC platform gave us access to productivity applications like Word and Excel and eventually to web applications like search and social networking. The mobile platform gave us messaging applications and location-based services. The ongoing democratization of ML will likely give us an amazing array of intelligent software and devices powering our world.

Highly scalable computing platforms

Even if specialized processors were available, not every company has the capital and skills needed to manage a large-scale computing platform needed to run advanced machine learning on a routine basis. This is where public cloud services such as Amazon Web Services (AWS), Google Cloud Platform, Microsoft Azure and others come in. These services offer developers a scalable infrastructure optimized for ML on rent and at a fraction of the cost of setting up on their own.
Open-source, deep-learning software frameworks

A major issue in the wide-scale adoption of machine learning is that there are many different software frameworks out there. Big companies are open sourcing their core ML frameworks and trying to push for some standardization. Just as the cost of developing mobile apps fell dramatically as iOS and Android emerged as the two dominant ecosystems, so too will machine learning become more accessible as tools and platforms standardize around a few frameworks. Some of the notable open source frameworks include Google’s TensorFlow, Amazon’s MXNet and Facebook’s Torch.
Developer-friendly tools.

The final step to democratization of machine learning will be the development of simple drag-and-drop frameworks accessible to those without doctorate degrees or deep data science training. Microsoft Azure ML Studio offers access to many sophisticated ML models through a simple graphical UI. Amazon and Google have rolled out similar software on their cloud platforms as well.
Marketplaces for ML algorithms and datasets.

Not only do we have an on-demand infrastructure needed to build and run ML algorithms, we even have marketplaces for the algorithms themselves. Need an algorithm for face recognition in images or to add color to black and white photographs? Marketplaces like Algorithmia let you download the algorithm of choice. Further, websites like Kaggle provide the massive datasets one needs to further train these algorithms.

“The final step to democratization of machine learning will be the development of simple drag-and-drop frameworks accessible.”

All of these changes mean that the world of machine learning is no longer restricted to university labs and corporate research centers that have access to massive training data and computing infrastructure.
What are the implications?

Back in the mid- and late-1990s, web development was done by specialists and was accessible only to firms with ample resources. Now, with simple tools like WordPress, Medium and Shopify, any lay person can have a presence on the web. The democratization of machine learning will have a similar impact of lowering entry barriers for individuals and startups.

Further, the emerging ecosystem, consisting of marketplaces for data, algorithms and computing infrastructure, will also make it easier for developers to pick up ML skills. The net result will be lower costs to train and hire talent. We think that the above two factors will be particularly powerful in vertical (industry-specific) use cases such as weather forecasting, healthcare/disease diagnostics, drug discovery and financial risk assessment that have been traditionally cost prohibitive.

Just like cloud computing ushered in the current explosion in startups, the ongoing build-out of machine learning platforms will likely power the next generation of consumer and business tools. The PC platform gave us access to productivity applications like Word and Excel and eventually to web applications like search and social networking. The mobile platform gave us messaging applications and location-based services. The ongoing democratization of ML will likely give us an amazing array of intelligent software and devices powering our world.

Market-based access to data and algorithms will lower entry barriers and lead to an explosion in new applications of AI. As recently as 2015, only large companies like Google, Amazon and Apple had access to the massive data and computing resources needed to train and launch sophisticated AI algorithms. Small startups and individuals simply didn’t have access and were effectively blocked out of the market. That changes now. The democratization of ML gives individuals and startups a chance to get their ideas off the ground and prove their concepts before raising the funds needed to scale.
But access to data is only one way in which ML is being democratized. There is an effort underway to standardize and improve access across all layers of the machine learning stack, including specialized chipsets, scalable computing platforms, software frameworks, tools and ML algorithms.

“Just like cloud computing ushered in the current explosion in startups … machine learning platforms will likely power the next generation of consumer and business tools.”

1. Specialized chipsets

Complex machine-learning algorithms require an incredible amount of computing power, both to train models and implement them in real time. Rather than using general-purpose processors that can handle all kinds of tasks, the focus has shifted towards building specialized hardware that is custom built for ML tasks. With Google’s Tensor Processing Unit (TPU) and NVIDIA’s DGX-1, we now have powerful hardware built specifically for machine learning.

Reproduced from Knowledge@Wharton

The Perilous World of Machine Learning for Fun and Profit 05-01

The Perilous World of Machine Learning for Fun and Profit: Pipeline Jungles and Hidden Feedback Loops

                                                                        George prototyping a machine learning model.

I haven't written a blog post in ages. And while I don't want to give anything away, the main reason I haven't been writing is that I've been too busy doing my day job at MailChimp. The data science team has been working closely with others at the company to do some fun things in the coming year.

That said, I got inspired to write a quick post by this excellent short paper out of Google,  "Machine Learning: The High Interest Credit Card of Technical Debt."

Anyone who plans on building production mathematical modeling systems for a living needs to keep a copy of that paper close.

And while I don't want to recap the whole paper here, I want to highlight some pieces of it that hit close to home.

Pipeline Jungles

Picture

George prototyping a machine learning model.

There was a time as a boy when my favorite book was George's Marvelous Medicine by Roald Dahl. The book is full of all that mischief and malice that makes Dahl books so much fun. 

In the book, George wanders around his house finding chemicals to mix up into a brown soup to give to his grandmother in place of her normal medicine. And reading this bit of felony grand-matricide as a child always made me smile.

Prototyping a new machine learning model is like George's quest for toxic chemicals. It's a chance for the data scientist to root around their company looking for data sources and engineering features that help predict an outcome.

A little bit of these log files. A dash of Google Analytics data. Some of Marge-from-Accounting's spreadsheet.

POOF! We have a marvelous model.

How fun it is to regale others with tales of how you found that a combination of reddit upvotes, the lunar calendar, and the number of times your yoga instructor says FODMAPs is actually somewhat predictive!

But now it's the job of some poor sucker dev to take your prototype model, which pulls from innumerable sources (hell, you probably scraped Twitter too just for good measure), and turn it into a production system.

All of a sudden there's a "pipeline jungle," a jumbled up stream of data sources and glue code for feature engineering and combination, to create something programmatically and reliably in production that you only had to create once manually in your George's-Marvelous-Medicine-revelry.

It's easy in the research and design phase of a machine learning project to over-engineer the product. Too many data sources, too many exotic and brittle features, and as a corollary, too complex a model. One trap the paper points out is leaving in low powered features in your prototype model, because well, they help a little, and they're not hurting anyone right? 

What's the value of those features versus the cost of leaving them in? That's extra code to maintain, maybe an extra source to pull from. And as the Google paper notes, the world changes, data changes, and every model feature is a potential risk for breaking everything.

Remember, the tech press (and vendors) would have you build a deep learning model that's fed scraped data from the internet's butthole, but it's important to exercise a little self-control. As the authors of the technical debt paper put it, "Research solutions that provide a tiny accuracy benefit at the cost of massive increases in system complexity are rarely wise practice." Preach.

Who's going to own this model and care for it and love it and feed it and put a band-aid on its cuts when its decision thresholds start to drift? Since it's going to cost money in terms of manpower and tied up resources to maintain this model, what is the worth of this model to the business? If it's not that important of a model (and by important, I'm usually talking top line revenue), then maybe a logistic regression with a few interactions will do you nicely.

Humans are Feedback Loop Machines

Picture

The graveyard at Haworth

In Haworth England, they used to bury bodies at the top of a hill above the town. When someone died, they got carted up to the overcrowded graveyard and then their choleraic juices would seep into the water supply and infect those down below, creating more bodies for the graveyard.

Haworth had a particularly nasty feedback loop.

Machine learning models suck up all sorts of nasty dead body water too.

At MailChimp, if I know that a user is going to be a spammer in the future, I can shut them down now using a machine learning model and a swift kick to the user's derriere.

But that future I'm changing will someday, maybe next week, maybe next year, be the machine learning system's present day.

And any attempt to train on present day data, data which has now been polluted by the business's model-driven actions (dead spammers buried at the top of the hill), is fraught with peril. It's a feedback loop. All of a sudden, maybe I don't have any spammers to train my ML model on, because I've shut them all down. And now my newly trained model thinks spamming is more unlikely than I know it to be.

Of course, such feedback loops can be mitigated in many ways. Holdout sets for example.

But we can only mitigate a feedback loop if we know about its existence, and we as humans are awesome at generating feedback loops and terrible at recognizing them. 

Think about time-travel in fiction. Once you have a time machine (and make no mistake, a well-suited ML model is pretty close to a forward-leaping time machine when it comes to things like sales and marketing), it's easy to jump through time and monkey with events, but it's hard to anticipate all the consequences of those changes and how they might alter your future training data.

And yet when the outputs of ML models are put in the hands of others to act on, you can bet that the future (and the future pool of training data with it) will be altered. That's the point! I don't predict spammers to do nothing about them! Predictions are meant to be acted upon.

And so, when the police predict that a community is full of criminals and then they start harassing that community, what do you think is going to happen? The future training data gets affected by the police's "special attention." Predictive modeling feeds back into systematic discrimination.

But we shouldn't expect cops to understand that they're burying their dead at the top of the hill.

This is one of my fears with the pedestrianization of data science techniques. As we put predictive models more and more in the hands of the layperson, have we considered that we might cut anyone out of the loop who even understands or cares about their misuse?

Get Integrated, Stay Alert

The technical debt paper makes this astute observation, "It’s worth noting that glue code and pipeline jungles are symptomatic of integration issues that may have a root cause in overly separated 'research' and 'engineering' roles."

This is absolutely true. When data scientists treat production implementation as a black box they shove their prototypes through and when engineers treat ML packages as black boxes they shove data pipelines through, problems abound.

Mathematical modelers need to stay close to engineers when building production data systems. Both need to keep each other in mind and keep the business in mind. The goal is not to use deep learning. The goal is not to program in Go. The goal is to create a system for the business that lives on. And in that context, accuracy, maintainability, sturdiness...they all hold equal weight.

So as a data scientist keep your stats buds close and your colleagues from other teams (engineers, MBAs, legal, ....) closer with the goal of getting work done together. It's the only way your models will survive past prototype.

View at the original source

How machine learning will fuel huge innovation over the next 5 years 03-02

How machine learning will fuel huge innovation over the next 5 years  

Machine learning is coming into a golden age, and with it we’re seeing an awakening of possibilities formerly reserved for science fiction.

Machine learning (ML) is a computer’s way of learning from examples, and it’s one of the most useful tools we have for the construction of artificial intelligence (AI). It begins with the design of an algorithm that learns from collected data, creating machines that in most cases become smarter as data volumes intensify.

We’ve seen a breakthrough in the field of ML in the last five years in part due to the recent wealth of big data streams provided from high-speed internet, cloud computing, and widespread smartphone usage, leading to the birth of the now popular “deep learning” algorithms. Heavily- used applications that have emerged with ML at their core include recommendation systems like those from Netflix and Amazon, face recognition technology as seen in Facebook, email spam filters like those from Google and Microsoft, and speech recognition systems such as Siri.

While the depth of advancement is unknown, what we can say with high certainty is that development in this field in the past five years will be nothing compared to what we’re going to see in the five years to come. Based on machine learning’s current state, here are four predictions of what we could see in the near future:

Image-Based Recognition: The technology for image and video-based recognition is on the horizon, and with it a whole new experience for users. Thanks to deep learning, we are now at the dawn of computers recognizing images, and the people and actions within them, with high accuracy based on the image alone and with minimum reliance on external data. It’s not just new pictures that will become recognizable either, but the entire history of digitized images and video footage. This will massively change how these assets are located and shared online. For example, YouTube might soon intelligently find content related to parts of a clip you watched and liked based only on the visual content of the video itself. The resulting efficiencies in both our work and personal time will be profound.

Healthcare: Machine learning’s ability to analyze and store massive amounts of data should provide physicians with much-needed second opinions and lead to the detection and treatment of medical ailments on a mass scale. Packaged as smart, wearable computing devices, personal health monitors that detect various conditions as they arise should become widespread in the next five years, in a similar fashion to activity trackers like Fitbit. The advancements here could significantly accelerate our human desire to protect our own longevity and create major breakthroughs for the operations of the medical industry.

Travel & Communication: By 2020, real-time translation technology may be fully accessible. We’ll see everything from an app on your phone that instantly translates foreign signs and texts to phone conversations that are immediately converted to a listener’s native language, without speakers even knowing the difference. As globalization booms, the language lines will soon be crossed. Business, in particular, stands to benefit enormously from the advancement here, with tech giants such as Google and Microsoft already taking the necessary steps to build such tools, making the need for a premium multilingual workforce obsolete.

Advertising: Based on recent ML advancements, in just a few short years augmented reality technology should become the commonplace method for integrated branding. This will allow advertisers to seamlessly place products into existing content by properly identifying the depth, relative size, lighting, and shading of the product in comparison to the setting. This essentially makes any historical video property available for integration. The computer vision technology firm Mirriad has already been heralded (and won an Oscar) for its advancements in the field. Looking at online video, as companies continue to try and tap into hugely popular amateur content, this technology will revolutionize their capabilities.

So while we have already seen enormous advancements in the fields above of late, a full-scale commercialization of machine learning technologies could be seen as soon as 2020. While I’ve only listed a few predictions above, almost all sectors of the economy stand to benefit enormously from the efficiencies of this new era of machine learning. We are already seeing a swell in consumer demand in experiences that require ML at their core, and the examples above only touch the surface of what is possible. If things continue on the trajectory we expect, the golden age of machine learning might very well make the next five years in technology the most exciting yet.

View at the original source

Your Data Should Be Faster, Not Just Bigger 02-08

Your Data Should Be Faster, Not Just Bigger

It’s universally acknowledged that Big Data is now a fact of life, but while large enterprises have spent heavily on managing largevolumes and disparate varieties of data for analytical purposes, they have devoted far less to managing high velocity data. That’s a problem, because high velocity data provides the basis for real-time interaction and often serves as an early-warning system for potential problems and systemic malfunctions.

Moreover, data proliferation has been accelerating. EMC recently reported that data volumes can be expected to double every two years, with the greatest growth coming from the vast amounts of new data being produced by intelligent devices and sensors. Oracle president Mark Hurd has predicted that the number of devices connected to the Internet will grow from 9 billion to 50 billion by the end of this decade.

What makes device data, sensor data, and other forms of “fast data” distinctive is that, unlike historical data, it is live, interactive, automatically generated, and often self-correcting. Historical data is used to identify patterns that inform future decision-making, while fast data is designed for real-time decisions and real-time responses. Think of fast data as the continuous processing of events and data in order to gain instantaneous insight and take instantaneous action.

While fast data is not really new, it has been largely restricted to a couple of high-value uses: complex event processing (CEP) activities that operate on event streams, examples of which include algorithmic trading and fraud monitoring in financial services; and event correlation, which includes the systems that monitor and manage complex industrial components such as jet engines.

So, what has changed? First, the explosion of fast data has driven the demand for instant action. Second, innovators in social media and services like Uber have shown that businesses can be differentiated based on their ability to act on data instantly. Uber knows where you are, where you’re going, and how you will pay to get there because it can capture, analyze, and act on data in real time. The availability of lower-cost memory is making fast data accessible for a broader set of applications, including:

• First responder systems that rely on integrating fast response data collection and analysis
• Network usage systems that respond instantly based on traffic patterns
• Customer-experience management systems that analyze vast amount of behavioral data in real time to tailor interactions and support self-service

Organizations that know how to use fast data will be more nimble, adaptive, and competitive. What can companies do differently to prepare for the opportunities created by high-velocity data? Here are a couple of suggestions:

1. Automate decision-making to increase customer engagement. Monitor customer activities to identify—and respond to—patterns, thresholds, and triggers. One major retailer is seeking to engage with customers in real time while they are online, but is hampered by traditional systems and batch processing environments. They are now creating an environment that marries customer and inventory data with streaming data so they can, for example, report to the customer whether a product is in stock and address any inventory gaps immediately.
2. Integrate machine-generated data to personalize interactions.EMC projects that soon nearly two-thirds of all data will be generated by machines, not people. That presents a technical challenge: how can companies capture and analyze many flows of data concurrently? The good news is that next-generation data systems that prioritize real-time data are in the early stages of adoption. By integrating new sources of machine-generated data, and combining it with traditional data sources, firms can further personalize their customer interactions. Ericsson, the mobile broadband company, has developed real-time visibility into its system performance, using device and user data as it is generated. This enables Ericsson to identify and improve slow performance as needed and to serve customers with programming tailored to them.

Firms that take these steps now will be well positioned to improve their operations and better serve their customers using high-speed data.

View at the original source

Follow The Data Down The Rabbit Hole 02-03

Follow The Data Down The Rabbit Hole

When it comes to analyzing big data, often we are urged to “follow the data,” “unlock its secrets,” analyze it and expect obscure truths to finally be revealed. While this is essentially true, data can tell many tales, and the way analytics are used nowadays, those tales greatly differ, depending on the human minds that interpret them.

Who asks the questions? 

The question of human bias hangs like a shadow over the accuracy and efficiency of big data analytics, and thus the viability of answers obtained thereof. If different humans can look at the same data and come to different conclusions, just how reliable can those deductions be?

There is no question that using data science to extract knowledge from raw data provides tremendous value and opportunity to organizations in any sector, but the way it is analyzed has crucial bearing on that value.

In order to extract meaningful answers from big data, data scientists must decide which questions to ask of the algorithms. However, as long as humans are the ones asking the questions, they will forever introduce unintentional bias into the equation. Furthermore, the data scientists in charge of choosing the queries are often much less equipped to formulate the “right questions” than the organization’s specialized domain experts.

For example, a compliance manager would ask much better questions about her area than a scientist who has no idea what her day-to-day work entails. The same goes for a CISO or the executive in charge of insider threats. Does this mean that your data team will have to involve more people all the time? And what happens if one of those people leaves the company?

Data science is necessary and important, and as data grows, so does the need for experienced data scientists. But at the same time, leaving all the computational work to humans makes it slower, less scientific, and quick to degrade in quality because the human mind cannot keep up with the quantum leap that big data is undergoing.

The scalability issue

Scalability is an urgent problem for big data and data science that is growing rapidly. According toresearch by MGI and McKinsey: “The United States alone faces a shortage of 140,000 to 190,000 people with analytical expertise and 1.5 million managers and analysts with the skills to understand and make decisions based on the analysis of Big Data.” 

Data scientists are already in very short supply, while the amounts of data that organizations generate and wish to leverage only grow as every industry from healthcare to critical infrastructure looks to big data to help them accelerate business and solve problems.

A joint research study by GE and Accenture states that “80-90% of companies across the industries surveyed indicated that big data analytics is either the top priority for the company or in the top three.” Furthermore, “53% of senior executives of industrial companies around the world say Big Data analytics is now a board level initiative.” At this rate, can enough data scientists become qualified and experienced quickly enough to respond to the galloping needs companies already have for real-time analytics? Clearly not. Scalability is already a major issue that needs to be solved—fast.

Luckily, a solution to this issue is already within reach.

Machine learning 

Rather than requiring data scientists to analyze and query big data, a wiser and more efficient way to maximize its benefit is to leave the detection phase to machine learning. Data scientists would then merely have to examine and classify the results – anomalies, events and issues that can only be discerned by human eyes.

Due to technological breakthroughs that are already available today, highly sophisticated analytic algorithms can automatically detect or predict problems by instantaneously analyzing unlimited amounts of very complex data without bias, time commitment or excessive false positives.

The Holy Grail here is automating analytics in a way that is reliable, accurate and relevant to different organizational needs, without necessitating any manual intervention. Solutions that offer this type of no-fuss big-data capabilities can reduce or eliminate the detection work required of data scientists, allowing them to focus on results. This in turn reduces costs for the organization and minimizes long-term spending on the ongoing deployment of any big-data solution it has in place.

Ultimately, by solving the issues that prevent full optimization of big data analytics — especially the human factor and its disproportionate impact on the current-day process — organizations will be able to detect and address all types of threats and opportunities much more rapidly. This capability is becoming increasingly crucial in an era when data is being generated by both humans and machines, and is sure to become a pivotal way for businesses to create situational awareness, detect issues and optimize operations to achieve their business objectives.

View at the original source