In this paper, we propose a method for home activity monitoring. We demonstrate our model on dataset of Detection and Classification of Acoustic Scenes and Events (DCASE) 2018 Challenge Task 5. This task aims to classify multi-channel audios into one of the provided pre-defined classes. All of these classes are daily activities performed in a home environment. To tackle this task, we propose a gated convolutional neural network with segment-level attention mechanism (SAM-GCNN). The proposed framework is a convolutional model with two auxiliary modules: a gated convolutional neural network and a segment-level attention mechanism. Furthermore, we adopted model ensemble to enhance the capability of generalization of our model. We evaluated our work on the development dataset of DCASE 2018 Task 5 and achieved competitive performance, with a macro-averaged F-1 score increasing from 83.76% to 89.33%, compared with the convolutional baseline system.

Artificial perception is traditionally handled by hand-designing task specific algorithms. However, a truly autonomous robot should develop perceptive abilities on its own, by interacting with its environment, and adapting to new situations. The sensorimotor contingencies theory proposes to ground the development of those perceptive abilities in the way the agent can actively transform its sensory inputs. We propose a sensorimotor approach, inspired by this theory, in which the agent explores the world and discovers its properties by capturing the sensorimotor regularities they induce. This work presents an application of this approach to the discovery of a so-called visual field as the set of regularities that a visual sensor imposes on a naive agent's experience. A formalism is proposed to describe how those regularities can be captured in a sensorimotor predictive model. Finally, the approach is evaluated on a simulated system coarsely inspired from the human retina. Keywords: autonomous systems, developmental robotics, sensorimotor contingencies, predictive processing, sensorimotor learning, humanlike vision 1. Introduction Autonomy in robotics relies on sensory data processing to capture information about the world and adapt to it. Although the influence of machine learning has been growing more important in the last decades, traditional approaches to this problem of data processing involve significant manual design from engineers that build the robot. Consequently the resulting techniques for artificial perception appear rigid and constrained for tractability. Each of these specialized algorithms is applicable to only a small set of tasks, with potentially limiting inbuilt biases from the designer. Corresponding author Email address: alaflaquiere@aldebaran.com (Alban Laflaquière) Preprint submitted to Neurocomputing October 5, 2018 autonomy in a robot. Instead, an autonomous robot must be able to cope with the complexity of its world, build its own way to perceive it and adapt to its variations. To address this issue, the field of developmental robotics takes inspiration from biological and cognitive development in children [4]. It proposes that an agent learns to interact with its environment, autonomously and on an ontogenic timescale.

We construct a general unified framework for learning representation of structured data, i.e. data which cannot be represented as the fixed-length vectors (e.g. sets, graphs, texts or images of varying sizes). The key factor is played by an intermediate network called SAN (Set Aggregating Network), which maps a structured object to a fixed length vector in a high dimensional latent space. Our main theoretical result shows that for sufficiently large dimension of the latent space, SAN is capable of learning a unique representation for every input example. Experiments demonstrate that replacing pooling operation by SAN in convolutional networks leads to better results in classifying images with different sizes. Moreover, its direct application to text and graph data allows to obtain results close to SOTA, by simpler networks with smaller number of parameters than competitive models.

In this work, we attempt to answer a critical question: whether there exists some input sequence that will cause a well-trained discrete-space neural network sequence-to-sequence (seq2seq) model to generate egregious outputs (aggressive, malicious, attacking, etc.). And if such inputs exist, how to find them efficiently. We adopt an empirical methodology, in which we first create lists of egregious output sequences, and then design a discrete optimization algorithm to find input sequences that will cause the model to generate them. Moreover, the optimization algorithm is enhanced for large vocabulary search and constrained to search for input sequences that are likely to be input by real-world users. In our experiments, we apply this approach to dialogue response generation models trained on three real-world dialogue data-sets: Ubuntu, Switchboard and OpenSubtitles, testing whether the model can generate malicious responses. We demonstrate that given the trigger inputs our algorithm finds, a significant number of malicious sentences are assigned large probability by the model, which reveals an undesirable consequence of standard seq2seq training. Recently, research on adversarial attacks (Goodfellow et al., 2014; Szegedy et al., 2013) has been gaining increasing attention: it has been found that for trained deep neural networks (DNNs), when an imperceptible perturbation is applied to the input, the output of the model can change significantly (from correct to incorrect). This line of research has serious implications for our understanding of deep learning models and how we can apply them securely in real-world applications. It has also motivated researchers to design new models or training procedures (Madry et al., 2017), to make the model more robust to those attacks.

And you thought Uber Black cars were nice. Renault has revealed its vision for a self-driving car at the Paris Motor Show. It's called the EZ-Ultimo, and the flamboyant, neo-retro sedan features enormous sliding doors and a lounge-like interior with nothing that even resembles a driver's seat. The very long and low-riding electric car can automatically lift itself over bumps, and uses four-wheel steering to navigate the sort of tight European city streets it was designed to cruise around. The automaker says it has Level 4 autonomy, which means it can operate within a pre-mapped area, so it's not for going off the beaten path.

Unlike recipients James P. Allison and Tasuku Honjo, most of us are unlikely to accept such an award in Stockholm. Neither are our beloved pet dogs, even though owners will swear their four-legged companions are geniuses. Sure, dogs are smart--at least when it comes to working with humans. But pigs, for instance, are smarter than you think. That's the contention of a new paper out today in the journal Learning & Behavior, which asks, "in what sense are dogs special?"

"This is when I start feeling my age," says Anne Corcoran. Corcoran leads a group that looks at how our genomes – the DNA coiled in almost every cell in our bodies – relate to our immune systems, and specifically to the antibodies we make to defend against infection. She is, in her own words, an "old-school biologist," brought up on the skills of pipettes and Petri dishes and protective goggles, the science of experiments with glassware on benches – what's known as "wet lab" work. "I knew what a gene looked like on a gel," she says, thinking back to her early career. These days that skill set is not enough. "When I started hiring Ph.D. students 15 years ago, they were entirely wet lab," Corcoran says. "Now when we recruit them, the first thing we look for is if they can cope with complex bioinformatic analysis." To be a biologist, nowadays, you need to be a statistician, or even a programmer.

As whispers of another UK election grow louder, Tom Watson, Deputy Leader of the Labour Party and Shadow Secretary for Digital, Culture, Media, and Sport, could soon play a pivotal role in the future of AI in Britain. In this world exclusive interview for AI Business, Watson outlines his views on how a Labour government could use AI and automation to address inequality and the future of public services, as well as deliver improved working and living conditions for all. LONDON--Tom Watson has been given many labels during the course of his 20-year career as a Labour politician. Since calling for Tony Blair's resignation back in 2007, he gained a reputation among some as Labour's'arch-fixer'. Today, as deputy leader of the Labour Party, he's often seen by some on the left (ironically) as a'Blairite' opponent of the party's leader, Jeremy Corbyn.

CLAIRE is an initiative by the European AI community that seeks to strengthen European excellence in AI research and innovation. To achieve this, CLAIRE proposes the establishment of a pan-European Confederation of Laboratories for Artificial Intelligence Research in Europe that achieves "brand recognition" similar to CERN. We believe that artificial intelligence (AI) will fundamentally change the way we live and work. It is also likely to become crucial in addressing society's grand challenges. In addition, AI is a global "game changer" that has become a major driver of innovation, future growth and competitiveness.

Artificial intelligence is transforming the traditional delivery of legal services. In general terms, the set of tools broadly called "legal analytics" promises to do two things: increase the efficiency of tasks that once required substantial time and human effort, and mine masses of data to discover new insights that were previously inaccessible. As legal scholars, we're excited about the promise of applying these tools to legal research questions. Students are involved too, so that we can educate the next generation of lawyers to leverage these tools in their own practices. Suppose that a company wants to forecast which employee complaints lead to lawsuits.