In this work, we present a methodology that enables an agent to make efficient use of its exploratory actions by autonomously identifying possible objectives in its environment and learning them in parallel. The identification of objectives is achieved using an online and unsupervised adaptive clustering algorithm. The identified objectives are learned (at least partially) in parallel using Q-learning. Using a simulated agent and environment, it is shown that the converged or partially converged value function weights resulting from off-policy learning can be used to accumulate knowledge about multiple objectives without any additional exploration. We claim that the proposed approach could be useful in scenarios where the objectives are initially unknown or in real world scenarios where exploration is typically a time and energy intensive process. The implications and possible extensions of this work are also briefly discussed.

Conventional text classification models make a bag-of-words assumption reducing text, fundamentally a sequence of words, into word occurrence counts per document. Recent algorithms such as word2vec and fastText are capable of learning semantic meaning and similarity between words in an entirely unsupervised manner using a contextual window and doing so much faster than previous methods. Each word is represented as a vector such that similar meaning words such as 'strong' and 'powerful' are in the same general Euclidian space. Open questions about these embeddings include their usefulness across classification tasks and the optimal set of documents to build the embeddings. In this work, we demonstrate the usefulness of embeddings for improving the state of the art in classification for our tasks and demonstrate that specific word embeddings built in the domain and for the tasks can improve performance over general word embeddings (learnt on news articles, Wikipedia or PubMed).

Linking human whole-body motion and natural language is of great interest for the generation of semantic representations of observed human behaviors as well as for the generation of robot behaviors based on natural language input. While there has been a large body of research in this area, most approaches that exist today require a symbolic representation of motions (e.g. in the form of motion primitives), which have to be defined a-priori or require complex segmentation algorithms. In contrast, recent advances in the field of neural networks and especially deep learning have demonstrated that sub-symbolic representations that can be learned end-to-end usually outperform more traditional approaches, for applications such as machine translation. In this paper we propose a generative model that learns a bidirectional mapping between human whole-body motion and natural language using deep recurrent neural networks (RNNs) and sequence-to-sequence learning. Our approach does not require any segmentation or manual feature engineering and learns a distributed representation, which is shared for all motions and descriptions. We evaluate our approach on 2,846 human whole-body motions and 6,187 natural language descriptions thereof from the KIT Motion-Language Dataset. Our results clearly demonstrate the effectiveness of the proposed model: We show that our model generates a wide variety of realistic motions only from descriptions thereof in form of a single sentence. Conversely, our model is also capable of generating correct and detailed natural language descriptions from human motions.

New research from Rocket Fuel on consumer perceptions of artificial intelligence (AI) shows younger generations welcome suggestions and predictions for products and services. Millennials are receptive to the use of AI in advertising, according to'Consumer Perceptions of AI' survey results from predictive marketing company Rocket Fuel. The survey found that over two-thirds of millennials see the benefits of brands using AI to help inform and direct their buying decisions. "Our research provides a snapshot of consumer attitudes towards AI and the fast-changing digital landscape," says Mailee Creacy, country manager at Rocket Fuel ANZ. "This is especially true for Millennials who are aware of the value exchange that takes place – they provide brands with personal data and expect to see their information used in ways that provides them with tangible benefits. Being able to engage with Millennials in a personalised way is the next frontier as brands seek to maintain and increase relevancy in the digital age," she says.

Japanese furniture maker France Bed Co. said Tuesday that it has developed a robot baby for people with dementia. The realistic-looking robot cries and laughs like a real baby when its hands and feet are touched. The product is expected to invigorate patients' emotional expression and sparks communication, according to the subsidiary of France Bed Holdings Co. The robot is 47 cm tall and weighs 1.4 g. It will sell for ¥15,984 at stores dealing in nursing care goods from May 25.

If you could have any superpower, which would it be? For Hilary Mason, it's not a question of if, because she already has one: She uses technology to unearth hidden insights, from how to catch people's attention online, to where you can find the best burgers in New York City. She has been at the forefront of machine learning, AI, and analytics. Hilary will discuss how your business can use these tools to move from insight to action to competitive advantage faster than a speeding bullet. She is the co-founder of Fast Forward Labs, Data Scientist in Residence at Accel and is the former Chief Scientist at Bitly.

At the Think About! conference staged in Warsaw by Ringier Axel Springer, the eastern Europe joint venture of Ringier and Axel Springer, I talked about the changes that AI, artificial intelligence, will bring to the profession of journalism in the future. You can already find AI in a huge amount of online products, for example in thematic pages that are almost completely automatically generated. The articles appear in a narrow thematic context, such as political or cultural content, but the possibilities here are almost endless. And with personalized content from online content providers, such as Axel Springer's news platform for Samsung smartphones, UPDAY, an algorithm learns from the user which issues and format he or she prefers, and constantly improves the personal selection of this content.

Propensity score analysis is used when experimentation is not feasible or as a recourse when experiments go awry ("broken" experiments). Its basic concepts were hammered out over the span of several decades by Jerzy Neyman, William Cochrane, Donald Rubin and several other eminent statisticians, and the thinking of distinguished economist James Heckman has also influenced its development. Propensity score analysis in several variations has seen extensive use in medical research, economics, education, assessment of government programs and, more recently, in marketing research and predictive analytics. First, why do we use experiments? We may wish to test the efficacy of some treatment or intervention such as medication, therapy and counseling or, in the case of marketing, liking for a new product.

If you've spent any time barking at your virtual assistant, you've no doubt had the conversation run aground a few times. "No, Siri, I said'Play Prince Purple Rain,' not'Belay price Urkle T-pain.'" But while Siri, Alexa, and Google may sometimes have a tough time understanding what you're asking even if you speak in a plain American accent, just imagine what it's like for somebody speaking English with a foreign accent. We put the three top assistants to the test. We asked friends of ours from Italy, Ireland, Scotland, England, Japan, Germany, and Australia to ask an iPhone with Siri, a Google Home, and an Amazon Echo various questions.

Sometimes when it rains, it pours, including official confirmation of new Far Cry, Assassin's Creed and The Crew games in the offing, thanks to Ubisoft doing what amounts to a celebratory dance after an upbeat fiscal earnings report. The France-based multinational games conglomerate isn't getting specific yet -- we'll likely have to wait until E3 mid-June to learn more. But it posted teaser images for Far Cry 5, The Crew 2, and an untitled Assassin's Creed game by way of the latter franchise's symbol above the words "A new era begins." Far Cry is the company's sandbox pulp-adventure series, usually transpiring in perilous renditions of remote tropical islands or romanticized non-Western locales. The Crew, which debuted in in 2014, is that same open-ended approach mapped onto a mass driving sim with automotive roleplaying elements.