EdgeTier, an Irish start-up using artificial intelligence (AI) to improve customer service, has raised €1.5 million in a new funding round. The investment is to be used to to increase research and development efforts in its AI assistant known as Arthur, and for further international expansion. Arthur uses AI, analytics and automation to guide call centre staff through complex customer queries, resulting in speedier and more accurate responses. Founded in 2015 by Shane Lynn, Bert Lehane and Ciarán Tobin, the company recently edged out rival start-ups Webio, Data Chemist, VRAI and UrbanFox to win Enterprise Ireland's'digital disruptor' award. Last year the company also won'best start-up' at AI Ireland's awards and'best use of data science in a start-up' at the DatSci awards.

Google officials have admitted that the company's workers can listen to Google Assistant users, and that one of them recently leaked confidential data. A Dutch language expert working for Google to train its speech technology leaked private information in a breach of the company's security policies, company officials said. The disclosure came after Belgian broadcaster VRT NWS reported that its reporters listened to more than 1,000 conversations recorded by the search giant's virtual assistant, including some that revealed identifiable information about the users. "As part of our work to develop speech technology for more languages, we partner with language experts around the world who understand the nuances and accents of a specific language," Google executive David Monsees wrote in a blog post posted on Thursday. These language experts review and transcribe a small set of queries to help us better understand those languages." "We just learned that one of these language reviewers has violated our data security policies by leaking confidential Dutch audio data.

Elon Musk's brain-computing venture Neuralink has scheduled an event on Tuesday to "share" what it has been secretively working on since its start a few years ago. Musk, who is also the chief executive of Tesla Inc. TSLA, 2.72% and privately held Space Exploration Technologies Corp. and The Boring Co., late Thursday retweeted Neuralink's announcement: We're having an event next Tuesday in San Francisco to share a bit about what we've been working on the last two years, and we've reserved a few seats for the internet. Neuralink's bare-bones website has job postings that describe the company as developing "ultra-high bandwidth brain-machine interfaces to connect humans and computers" and currently building a team of "multidisciplinary experts." Related: Tesla picks up two'car of the year' awards Most if not all posts requested future employees who were "resourceful, flexible and adaptable; no task is too big or too small." Musk confirmed plans for Neuralink in 2017 and last year, during his infamous appearance on The Joe Rogan Experience's podcast, said he'd be ready to announce a new Neuralink product in a few months, and that "a best-case scenario" would be humans effectively merging with artificial intelligence.

Artificial intelligence (AI) is set to play a key role in the future of financial services and more broadly in what UBS and the World Economic Forum refer to as the "Fourth Industrial Revolution." The global economy is on the cusp of profound changes driven by "extreme automation" and "extreme connectivity." In this changing economic landscape, AI is expected to be a pervasive feature, allowing to automate some of the skills that formerly only humans possessed. In the financial services industry in particular, there has been a lot of noise around the potential of AI and data supports that investors are excited about the impact the technology could have across the industry. VC-backed fintech AI companies raised approximately US$2.22 billion in funding in 2018, nearly twice as much as 2017's record.

As a cucumber plant grows, it sprouts tightly coiled tendrils that seek out supports in order to pull the plant upward. This ensures the plant receives as much sunlight exposure as possible. Now, researchers at MIT have found a way to imitate this coiling-and-pulling mechanism to produce contracting fibers that could be used as artificial muscles for robots, prosthetic limbs, or other mechanical and biomedical applications. While many different approaches have been used for creating artificial muscles, including hydraulic systems, servo motors, shape-memory metals, and polymers that respond to stimuli, they all have limitations, including high weight or slow response times. The new fiber-based system, by contrast, is extremely lightweight and can respond very quickly, the researchers say.

Artificial intelligence (AI) is one of the prime technologies leading the wave of disruption that is going on within the health care sector. Recent studies have shown that AI technology can outperform doctors when it comes to cancer screenings and disease diagnoses. In particular, this could mean specialists such as radiologists and pathologists could be replaced by AI technology. Per an article by the Association of American Medical Colleges, "a New England Journal of Medicine article predicted that'machine learning will displace much of the work of radiologists and anatomical pathologists,' adding that'it will soon exceed human accuracy.' That same year, Geoffrey Hinton, PhD, a professor emeritus at the University of Toronto who also designs machine learning algorithms for Google (and who received the Association for Computing Machinery's A.M. Turing Award, often called the Nobel Prize of computing, in 2019), declared, 'We should stop training radiologists now.'"

Building on the view of machine learning as search, we demonstrate the necessity of bias in learning, quantifying the role of bias (measured relative to a collection of possible datasets, or more generally, information resources) in increasing the probability of success. For a given degree of bias towards a fixed target, we show that the proportion of favorable information resources is strictly bounded from above. Furthermore, we demonstrate that bias is a conserved quantity, such that no algorithm can be favorably biased towards many distinct targets simultaneously. Thus bias encodes trade-offs. The probability of success for a task can also be measured geometrically, as the angle of agreement between what holds for the actual task and what is assumed by the algorithm, represented in its bias. Lastly, finding a favorably biasing distribution over a fixed set of information resources is provably difficult, unless the set of resources itself is already favorable with respect to the given task and algorithm.

In this paper, we introduce the Preselection Bandit problem, in which the learner preselects a subset of arms (choice alternatives) for a user, which then chooses the final arm from this subset. The learner is not aware of the user's preferences, but can learn them from observed choices. In our concrete setting, we allow these choices to be stochastic and model the user's actions by means of the Plackett-Luce model. The learner's main task is to preselect subsets that eventually lead to highly preferred choices. To formalize this goal, we introduce a reasonable notion of regret and derive lower bounds on the expected regret. Moreover, we propose algorithms for which the upper bound on expected regret matches the lower bound up to a logarithmic term of the time horizon.

We study the problem of detecting adverse drug events in electronic healthcare records. The challenge in this work is to aggregate heterogeneous data types involving diagnosis codes, drug codes, as well as lab measurements. An earlier framework proposed for the same problem demonstrated promising predictive performance for the random forest classifier by using only lab measurements as data features. We extend this framework, by additionally including diagnosis and drug prescription codes, concurrently. In addition, we employ a recursive feature selection mechanism on top, that extracts the top-k most important features. Our experimental evaluation on five medical datasets of adverse drug events and six different classifiers, suggests that the integration of these additional features provides substantial and statistically significant improvements in terms of AUC, while employing medically relevant features.

Stochastic Gradient Descent (SGD) is one of the workhorses of modern day machine learning. In many nonconvex optimization problems, such as training deep neural networks, SGD is able to produce solutions with good generalization error. Further, there is evidence that the generalization error of an SGD solution can be significantly better than Gradient Descent (GD) [12]. This suggests that, to understand the behavior of SGD, it is not enough to consider the limiting cases (such as small step-size or large batch-size), when it degenerates to GD. We take an alternate view of SGD as a sampling algorithm, and aim to understand its convergence to an appropriate stationary distribution.