Cognitive Computing Powers 6 Smart Deployments
Source: Geralt
Cognitive computing is going to transform a lot of things. Among them, how humans and machines interact, how companies interface with their customers, and how cybersecurity threats are managed. In fact, its potential uses are practically limitless, although not everyone is approaching the problem in the same way. Regardless of which technologies innovators are using or planning to use, their common goal is to mimic human reasoning at machine-level scale and velocity.
Sponsor video, mouseover for sound
Since IBM Watson won a Jeopardy! match on television in 2011, IBM commercialized its technology, expanded its set of APIs and launched a new consulting practice. In September 2015, the IBM Watson team announced it had more than 100 partners developing cognitive apps, products, and services going into market. However, IBM Watson is just one example of how companies are approaching cognitive computing.
More companies are using machine learning, artificial intelligence (AI), natural language processing (NLP), or some combination of those and other technologies to make better sense of the vast volume, variety, and velocity of big data.
Some of them are adding or plan to add more types of technology to the mix, such as contextual awareness, neural networks, more sophisticated pattern recognition, visual sensing, and other technologies that, when combined, enable machines to more accurately emulate human thought and reasoning. In short, cognitive computing is a superset of technologies that are being combined in different ways to solve both general and specific problems.
However, the reasoning capabilities, matching capabilities, learning capabilities, and language-processing capabilities that will enable organizations to uncover new levels of insights may well raise a new wave of privacy and security concerns along the way.
These technologies will likely fuel the debate about the balance between humans and machines as they gain momentum, whether regarding the fear of replacing more human decision-making with automated systems or the doomsday AI threat some perceive.
On the other hand, there is a limit to what humans can do in today's real-time economy, which is why the seeds of cognitive computing are already sprouting across industries such as healthcare, financial services, and event management.
Page through this slideshow to learn how some of the capabilities are being used today.
Understand Your Audience
Performers, talent managers, and venue owners want to understand their audiences at a more personal level so they can fine-tune their content, improve audience experiences, order the right number and sizes for promotional items, and improve marketing ROI. Social aggregation tool provider Ampsy has developed a geofencing technology that aggregates, curates, and displays fan-generated content from live events. The company is using IBM Watson technology for sentiment analysis to understand how an entire audience feels about a concert generally, as well as specifics such as keywords and key phrases and how the audience felt about the playlist, the service, the security, and the parking.
"The venue owner, band manager, event producer, and talent [can understand] what fans are feeling in the live moment," said Jeremy Gocke, founder and CEO of Ampsy. "The stakeholders want to understand who their fans are and have the ability to micro-market to them."
A client can take the content gathered from its geofence and embed it on its website or on a social property and encourage fans to use a certain hashtag when sharing their content. There is also a personality-insights option that can analyze a fan's historical tweets to derive a persona. A manufacturer is using Ampsy to understand sentiment about the band merchandise it produces.
Comprehend Emotion
Predicting emotion isn't easy because how a person feels, what they perceive, and the decisions they make can change based on context. Wearables and IoT strategy and execution agency Amyx+McKinsey is developing technology that can understand customers in context -- where they are, what they're doing, who they're with, and how they're feeling in the moment -- using data gathered from smartphones, smartwatches, and the IoT.
"We think cognitive science will be popular on the retail side. When you walk into a store, they have lots of data about their product and about you, but they don't know how you're going to behave right then and there," said Scott Amyx, founder and CEO of Amyx+McKinsey. "Once we understand what's going on, what they're interested in the most, [we can] send those trigger points to the campaign management system, email, social media, and other chat forums and say you have an affinity with this particular product. Moreover, your emotion unlocks an opportunity which might be a promotion or a coupon."
Improve Investment Performance
Investment managers, asset managers, and hedge fund managers need to understand the signals buried in mountains of content and be able to execute high-frequency trades. Their ability to outperform market benchmarks depends on their ability to comprehend the effect signals will have on investments or potential investments before their competitors do. Rage Frameworks uses cognitive intelligence, including computational linguistics and natural-language processing, to automatically interpret information from more than 200,000 sources globally as it arrives. The offering determines the relevance of content and the impact a signal will have on a security, company, stock, or bond.
"Active managers are overrun with the amount of data they have at their disposal. There's so much information -- so many things that impact companies like Apple -- that it's humanly impossible to do. This is a problem custom-made for cognitive computing," said Venkat Srinivasan, CEO and chairman of Rage Frameworks.
Improve Cyber-Security
As the battle between the black hats and white hats continues, companies are looking for better ways to minimize the risk of security threats that have not yet been identified. Masergy Communications, a global cloud network platform provider, is slowly moving toward cognitive computing. Its original platform used probabilistic math and physical models. Later, it incorporated machine learning. Now, the company also uses artificial intelligence, and it's in the process of embracing neural networks and ordered neural networks.
"Some people are using machine learning for virus detection or malware detection. That's a classification problem because you're deciding whether something is good or bad. If it's not detected, then it's undetectable for all your customers," said Masergy chief scientist Mike Stute.
Masergy still uses machine learning to detect what's normal in a network. With a neural network, tuning is a process of modifying a dataset and rescoring it again and again -- sometimes randomly. Ultimately, the constant retuning of an anomaly system changes its detection behavior in an adaptive and continuous way.
Personalize Shopping Experiences
Demographic marketing doesn't cut it in the Digital Age because customer expectations are high and their attention spans are short. Marketing platform provider Retention Science uses machine learning and artificial intelligence to help its customers market more effectively to their customers on a one-to-one basis. Marketers can personalize offers and optimize the timing of offers, such as sending an email within the time-frame in which the customer is most likely to open it.
"A few years ago, it was more about predicting what people are thinking or going to do, but now people want to know what to do with it," said Andrew Wagge, co-founder and CTO of Retention Science. "They want to identify the proper treatment for each one of their customers. We can analyze all the data points and bring signals together to do that. A lot of what we do involves reinforcement learning, so it's about continually updating the predictions, learning from what's going on, learning from what we do, and then trying to take a different course of action based on whatever happens."
Improve Multichannel Marketing
Timely multichannel marketing analysis is often frustrated by information silos, technical challenges, and uncertainty about the best course of action. Omnichannel marketing analytics platform provider Datorama uses machine learning algorithms to provide actionable recommendations across channels. Customers can simply upload their data without worrying about the structure of the dataset or metadata since the platform analyzes the structure of the dataset and adds metadata automatically.
The platform can identify the same entity from 200 different datasets. That capacity allows it to automatically normalize the data and create a semantic model. The machine-learning algorithm uses the newly loaded data, as well as what it has learned from every dataset that has ever been loaded into the system. Datorama also combines machine learning with artificial intelligence to find relationships within a given dataset and connections to other datasets. In addition, the company also uses natural-language processing and semantic analysis as part of its smart-data-mapping process.
"The Waterfall implementation of identifying data sources and having technical people implement ETL has a negative effect on marketing BI because of the amount of data sources and the dynamic nature of the data sources," said Efi Cohen, CTO and cofounder of Datorama. "We're providing a way to handle the exact same problem, and the ROI is positive."
| }
|