The most appreciative advancements in statistical AI, the ones with the most meaning and potential to improve data’s worth to the enterprise, are deep learning deployments of computer vision and natural language technologies.

The distinctions between these applications involve much more than image recognition versus that of speech or language. Horizontal computer vision use cases pertain to some aspects of inter-machine intelligence, e.g., scanning videos or production settings for anomalies and generating alerts to initiate automated procedures to address them.

Conversely, natural language technologies provide the most effective cognitive computing application for furthering human intelligence, decision making, and the action required to extract business value from such perceptivity.

While the utility derived from image recognition largely varies according to the vertical, the capability for machines to understand natural language—for humans to interact with databases in layperson’s terms across sources—strikes at the core of converting the unstructured data of language into informed action.

Few organizations, regardless of their industry, could not benefit from this capacity. The application of deep neural networks and other machine learning models for this universal use case presents the greatest win for the enterprise, resolves the issue of unstructured data, and is currently taking the form of the following capabilities:

♦ Natural language generation: According to Forrester, natural language generation systems (such as those associated with Alexa and conversational AI systems) leverage “a set of rules, templates, and machine learning to generate language in an emergent, real-time fashion.” Accomplished solutions in this space rely on basic precepts of deep learning to generate text for an array of use cases.

♦ Smart process automation: The impactof equipping bots and other means of process automation with algorithms from cognitive statistical models is unprecedented. Instead of simply implementingthe various steps necessary for workflows, such approaches can actually complete them by rendering decisions conventionally relegated to humans.

♦ Spontaneous question-answering: Answering sophisticated, ad hoc questions across data sources has always posed a challenge for machine intelligence options. When backed by deep learning techniques and other aspects of AI, organizations can overcome this obstacle to profit from any unstructured, written data they have.

No one can deny the merits of deploying cognitive computing to accelerate data preparation or make back-end processes easier. However, the aforementioned applications of natural language technologies shift that ease and expedience to the front end. They’re the means of directly empowering business users with the peerless predictions of deep learning and, more importantly, redirecting its business value from fringe use cases to those impacting mission-critical business processes.

Natural language generation

When applied to natural language technologies, deep learning’s chief value proposition is the capacity to issue predictions— with striking accuracy, in some cases—about language’s composition, significance, and intention. Models involving deep neural networks facilitate these advantages with a speed and facility far surpassing conventional, labor-intensive methods of doing so. According to AX Semantics CTO Robert Weissgraeber, “Neural networks, trained with deep learning, are used in the generation process for grammar prediction, such as finding the plural of ‘feature’ or ‘woman.’”

Natural language generation has swiftly become one of the most useful facets of natural language technologies. Both Gartner and Forrester have recently developed market reports monitoring its progress. More importantly, it’s also revamping BI by accompanying everything from visualizations to reports with natural language explanations. Perhaps even more significantly, natural language generation-powered systems have expanded from conversational AI applications to include “product descriptions, automated personalized personalized messaging like personalized emails, listing pages like the Yellow Pages, and select journalism applications like election reporting, sports reporting, and weather,” Weissgraeber noted.

Natural language generation’s rise can be partly attributed to its extension of natural language processing (which is transitioning from being trained by rules to being to being trained by machine learning models) to include responses. Specifically, natural language generation employs natural language processing components such as dependency parsing and named entity extraction to analyze what the user writes, and then creates hints for the user to make his configuration faster, Weissgraeber explained.

The principal components of natural language generation systems include:

♦ Data extraction and generation: Thistool chain handles what Weissgraebertermed “classic data processing.”

♦ Topic- and domain-dependent configurations: Natural language generationsystems rely on this component to analyzedata’s meaning.

♦ Word/phrase configurations: Theseconfigurations are used to select differentphrases based on the desired meaning.

♦ Textual management: These elementsbring the “text together, with grammarprediction, correct sentences, text length,and formatting,” Weissgraeber said.

When combined with a system push or API for delivery, these natural language generation characteristics utilize deep learning for stunningly sophisticated use cases. Forrester indicates that in finance, this “technology can review data from multiple sources, including external market conditions and a client’s investment goals and risk profile, to produce a personalized narrative for each of an advisor’s clients.”

The post Next-generation natural language technologies: The deep learning agenda appeared first on Artificial Intelligence.

Artificial intelligence (AI) seems to be taking over the world and is even helping us combat the ongoing coronavirus pandemic, but so far it has been a product of human supervision – we teach computers to see patterns, just like we teach children to read. However, researchers believe the future of AI depends on systems that are capable of learning on their own, without any supervision.

What is supervised learning?

When a parent points towards a dog and tells the baby to “Look at the doggie,” the child learns and understands what to call the furry four-legged friends. This is an example of supervised learning, as pointed out by New York Times. However, when the baby stands and stumbles, over and over again, before she learns how to walk, that is something else.

Computers and humans are quite similar when it comes to learning. Just as we learn mostly through observation or trial and error, computers also have to pass through the stage of supervised learning before they can reach the human-level of intelligence.

Even if a supervised learning system reads all the books in the world, it would still not be able to achieve human-level intelligence because a large chunk of our knowledge and expertise is not penned down.

Limitations of human supervision

Supervised learning comprises of feeding data, including images, audio, or text that is fed into computer algorithms, which teams machines to do what they do. However, this learning method has its restrictions.

“There is a limit to what you can apply supervised learning to today due to the fact that you need a lot of labeled data,” said Yann LeCun, an expert in the field of machine learning and artificial intelligence, and a recipient of the Turing Award, the equivalent of a Nobel Prize in computer science, in 2018. He is also the vice president and chief A.I. scientist at Facebook.

Although learning methods that are not dependent on such human intervention are less explored, they have been overshadowed by the success of supervised learning and its many practical applications in the real world, from self-driving cars to smart speakers. But supervised learning still can’t do many of the tasks that are simple enough even for a toddler.

Artificial intelligence that learns on its own

Therefore, scientists leading the charge of artificial intelligence research have shifted their focus to less-supervised learning methods in which the artificial intelligence develops a common sense or sorts and carries out tasks by learning on its own.

“There’s self-supervised and other related ideas, like reconstructing the input after forcing the model to a compact representation, predicting the future of a video or masking part of the input and trying to reconstruct it,” said Samy Bengio, a research scientist at Google.

Scientists are also exploring reinforcement learning, which requires very limited supervision and does not rely on data. This learning method, pioneered by University of Alberta’s Richard Sutton, follows a reward-driven learning mode, essentially like a dog performing a trick to earn a treat. The strategy has been developed to teach computer systems to learn new actions on their own.

All they need to do is set a goal, and a reinforcement learning system will try to achieve the said goal through trial and error until it is consistently receiving a reward. A more appropriate term for this future AI is “predictive learning,” which means that systems not only recognize patterns but also predict outcomes and choose a course of action autonomously.

For instance, if a self-supervised computer system “watches” millions of videos on YouTube, it will gather a representation of the world from the clips and when the machine is asked to perform a particular task, it can take action based on what it has learned from the videos – in other words, teach itself.

The post Scientists’ next AI agenda: Making machines learn ‘common sense’ and ‘teach’ themselves appeared first on Artificial Intelligence.