Artificial intelligence is quickly growing in importance in the'smart building' sector. Paul Skelton looks at the road ahead for a complex technology. When Mark Chung received an unexpectedly high $500 monthly electricity bill, he turned to his utility for help and answers. However, despite'smart' meters being installed in his home, they were no help. So Mark – an electrical engineer trained at Stanford University – took matters into his own hands.

One of the longest-lasting and most successful video-game franchises is the Doom series, launched in 1993 and still going strong with over 10 million copies sold. The game is a first-person shooter in which a space marine battles to survive against various demons and zombies. The game is notable because it pioneered 3-D graphics for PCs running MS-DOS, introduced networked multiplay, and even allowed players to create their own game levels. Indeed, large numbers of Doom levels--both official and player-created--are now freely available online, forming a formidable corpus for study and research. And that raises an interesting possibility. Is it possible to use this data to train a deep-learning algorithm to create its own levels of Doom that a human would find compelling?

One of the longest-lasting and most successful video-game franchises is the Doom series, launched in 1993 and still going strong with over 10 million copies sold. The game is a first-person shooter in which a space marine battles to survive against various demons and zombies. The game is notable because it pioneered 3-D graphics for PCs running MS-DOS, introduced networked multiplay, and even allowed players to create their own game levels. Indeed, large numbers of Doom levels--both official and player-created--are now freely available online, forming a formidable corpus for study and research. And that raises an interesting possibility.

The tech giants already know this and are investing in democratizing AI to make tools and services more widely available, but the user experience (UX) of machine learning is still overlooked. Companies can make massive improvements to machine learning-based applications even without access to the same levels of data or talent as the biggest players -- compensating for a lack of data by building a great UI (more on this later). When we focus on AI as a tool and recognize how crucial usability is to widespread adoption, we can see that there are opportunities to enhance existing AI in ways that have nothing to do with progress toward human-level machine intelligence or artificial general intelligence. While flashy projects like DeepMind and Google Brain are more likely to make headlines than Google's more mundane implementations of AI, such as search, the latter is a vastly more profitable business. According to a recent MarketWatch article, Google has "made a massive multibillion-dollar bet on AI and machine learning," a bet I believe is nicely hedged on the question of whether there'll be another "AI winter," a period of reduced interest in AI.

Advanced Driver Assistance Systems (ADAS) improve driving safety significantly. They alert drivers from unsafe traffic conditions when a dangerous maneuver appears. Traditional methods to predict driving maneuvers are mostly based on data-driven models alone. However, existing methods to understand the driver's intention remain an ongoing challenge due to a lack of intersection of human cognition and data analysis. To overcome this challenge, we propose a novel method that combines both the cognition-driven model and the data-driven model. We introduce a model named Cognitive Fusion-RNN (CF-RNN) which fuses the data inside the vehicle and the data outside the vehicle in a cognitive way. The CF-RNN model consists of two Long Short-Term Memory (LSTM) branches regulated by human reaction time. Experiments on the Brain4Cars benchmark dataset demonstrate that the proposed method outperforms previous methods and achieves state-of-the-art performance.

Stylistic dialogue response generation, with valuable applications in personality-based conversational agents, is a challenging task because the response needs to be fluent, contextually-relevant, as well as paralinguistically accurate. Moreover, parallel datasets for regular-to-stylistic pairs are usually unavailable. We present three weakly-supervised models that can generate diverse polite (or rude) dialogue responses without parallel data. Our late fusion model (Fusion) merges the decoder of an encoder-attention-decoder dialogue model with a language model trained on stand-alone polite utterances. Our label-fine-tuning (LFT) model prepends to each source sequence a politeness-score scaled label (predicted by our state-of-the-art politeness classifier) during training, and at test time is able to generate polite, neutral, and rude responses by simply scaling the label embedding by the corresponding score. Our reinforcement learning model (Polite-RL) encourages politeness generation by assigning rewards proportional to the politeness classifier score of the sampled response. We also present two retrieval-based polite dialogue model baselines. Human evaluation validates that while the Fusion and the retrieval-based models achieve politeness with poorer context-relevance, the LFT and Polite-RL models can produce significantly more polite responses without sacrificing dialogue quality.

Remote sensing lacks methods like the word vector representations and pre-trained networks that significantly boost performance across a wide range of natural language and computer vision tasks. To fill this gap, we introduce Tile2Vec, an unsupervised representation learning algorithm that extends the distributional hypothesis from natural language -- words appearing in similar contexts tend to have similar meanings -- to geospatial data. We demonstrate empirically that Tile2Vec learns semantically meaningful representations on three datasets. Our learned representations significantly improve performance in downstream classification tasks and similarly to word vectors, visual analogies can be obtained by simple arithmetic in the latent space.

Automatic understanding of human affect using visual signals is of great importance in everyday human-machine interactions. Appraising human emotional states, behaviors and reactions displayed in real-world settings, can be accomplished using latent continuous dimensions (e.g., the circumplex model of affect). Valence (i.e., how positive or negative is an emotion) and arousal (i.e., power of the activation of the emotion) constitute the most popular and effective affect representations. Nevertheless, the majority of collected datasets this far, although containing naturalistic emotional states, have been captured in highly controlled recording conditions. In this paper, we introduce the Aff-Wild benchmark for training and evaluating affect recognition algorithms. We also report on the results of the First Affect-in-the-wild Challenge (Aff-Wild Challenge) that was recently organized on the Aff-Wild database, and was the first ever challenge on the estimation of valence and arousal in-the-wild. Furthermore, we design and extensively train an end-to-end deep neural architecture which performs prediction of continuous emotion dimensions based on visual cues. The proposed deep learning architecture, AffWildNet, includes convolutional and recurrent neural network (CNN-RNN) layers, exploiting the invariant properties of convolutional features, while also modeling temporal dynamics that arise in human behavior via the recurrent layers. The AffWildNet produced state-of-the-art results on the Aff-Wild Challenge. We then exploit the AffWild database for learning features, which can be used as priors for achieving best performances both for dimensional, as well as categorical emotion recognition, using the RECOLA, AFEW-VA and EmotiW 2017 datasets, compared to all other methods designed for the same goal.

Understanding customer sentiments is of paramount importance in marketing strategies today. Not only will it give companies an insight as to how customers perceive their products and/or services, but it will also give them an idea on how to improve their offers. This paper attempts to understand the correlation of different variables in customer reviews on a women clothing e-commerce, and to classify each review whether it recommends the reviewed product or not and whether it consists of positive, negative, or neutral sentiment. To achieve these goals, we employed univariate and multivariate analyses on dataset features except for review titles and review texts, and we implemented a bidirectional recurrent neural network (RNN) with long-short term memory unit (LSTM) for recommendation and sentiment classification. Results have shown that a recommendation is a strong indicator of a positive sentiment score, and vice-versa. On the other hand, ratings in product reviews are fuzzy indicators of sentiment scores. We also found out that the bidirectional LSTM was able to reach an F1-score of 0.88 for recommendation classification, and 0.93 for sentiment classification.

Sequence-to-sequence attention-based models on subword units allow simple open-vocabulary end-to-end speech recognition. In this work, we show that such models can achieve competitive results on the Switchboard 300h and LibriSpeech 1000h tasks. In particular, we report the state-of-the-art word error rates (WER) of 3.54% on the dev-clean and 3.82% on the test-clean evaluation subsets of LibriSpeech. We introduce a new pretraining scheme by starting with a high time reduction factor and lowering it during training, which is crucial both for convergence and final performance. In some experiments, we also use an auxiliary CTC loss function to help the convergence. In addition, we train long short-term memory (LSTM) language models on subword units. By shallow fusion, we report up to 27% relative improvements in WER over the attention baseline without a language model.