Neural sequence models exhibit limited compositional generalization ability in semantic parsing tasks. Compositional generalization requires algebraic recombination, i.e., dynamically recombining structured expressions in a recursive manner. However, most previous studies mainly concentrate on recombining lexical units, which is an important but not sufficient part of algebraic recombination. In this paper, we propose LeAR, an end-to-end neural model to learn algebraic recombination for compositional generalization. The key insight is to model the semantic parsing task as a homomorphism between a latent syntactic algebra and a semantic algebra, thus encouraging algebraic recombination. Specifically, we learn two modules jointly: a Composer for producing latent syntax, and an Interpreter for assigning semantic operations. Experiments on two realistic and comprehensive compositional generalization benchmarks demonstrate the effectiveness of our model. The source code is publicly available at https://github.com/microsoft/ContextualSP.

Data augmentation, the artificial creation of training data for machine learning by transformations, is a widely studied research field across machine learning disciplines. While it is useful for increasing the generalization capabilities of a model, it can also address many other challenges and problems, from overcoming a limited amount of training data over regularizing the objective to limiting the amount data used to protect privacy. Based on a precise description of the goals and applications of data augmentation (C1) and a taxonomy for existing works (C2), this survey is concerned with data augmentation methods for textual classification and aims to achieve a concise and comprehensive overview for researchers and practitioners (C3). Derived from the taxonomy, we divided more than 100 methods into 12 different groupings and provide state-of-the-art references expounding which methods are highly promising (C4). Finally, research perspectives that may constitute a building block for future work are given (C5).

AI diagnostics startup MediCircle Health has recently introduced in India a rapid spectrometry-based test that employs machine learning and artificial intelligence to detect COVID-19. Spectral Instant Test (SpectraLIT) is a point-of-care diagnostic platform that performs spectral analysis to accurately and instantly determine if a spectral pattern of a virus from a nasal or mouthwash sample resembles SARS-CoV-2, the virus causing COVID-19. The test can deliver results "within seconds of its use", according to a press release by MediCircle. The company shared that the portable solution can be used for entry screening at various airports, malls, schools and other venues. It can also potentially enable secure and real-time reporting to health and other designated authorities.

Abbreviation is Artificial Intelligence on a digital globe background. A power plant that will run on "artificial intelligence" is about to get underway in West Africa. The joint venture between Swiss-based Xcell Security House and Finance and U.S.-based Beyond Limits will embed intelligence and awareness into the operations -- something that will create more efficiencies, greater productivity, and increased environmental protections. When ordinary people hear about artificial intelligence -- AI for short -- they immediately think about how machines will replace humans. But as the experts explained to this reporter, AI is meant to eliminate "mundane activities" so that those running heavy industrial operations can solve problems and improve performance, which translates into healthier bottom lines.

Over the last decade, the concept of "games as a service" has revolutionised the way the interactive entertainment industry works. From the subscriptions introduced by massively multiplayer online adventures such as World of Warcraft to the seasonal battle passes of current online shooters, we're seeing a huge amount of focus on games that can sustain a lucrative community of players over several years. But where does that leave more offbeat ideas and concepts that couldn't support years' worth of play? Where does it leave the single-player narrative adventure – the blockbusting genre that brought us titles such as Metal Gear Solid, Red Dead Redemption and Mass Effect? It's a genre Sony has supported through funding the studios that make games such as The Last of Us, Spider-Man and God of War.

The ability to convey relevant and diverse information is critical in multi-document summarization and yet remains elusive for neural seq-to-seq models whose outputs are often redundant and fail to correctly cover important details. In this work, we propose an attention mechanism which encourages greater focus on relevance and diversity. Attention weights are computed based on (proportional) probabilities given by Determinantal Point Processes (DPPs) defined on the set of content units to be summarized. DPPs have been successfully used in extractive summarisation, here we use them to select relevant and diverse content for neural abstractive summarisation. We integrate DPP-based attention with various seq-to-seq architectures ranging from CNNs to LSTMs, and Transformers. Experimental evaluation shows that our attention mechanism consistently improves summarization and delivers performance comparable with the state-of-the-art on the MultiNews dataset.

Extracting the interaction rules of biological agents from moving sequences pose challenges in various domains. Granger causality is a practical framework for analyzing the interactions from observed time-series data; however, this framework ignores the structures of the generative process in animal behaviors, which may lead to interpretational problems and sometimes erroneous assessments of causality. In this paper, we propose a new framework for learning Granger causality from multi-animal trajectories via augmented theory-based behavioral models with interpretable data-driven models. We adopt an approach for augmenting incomplete multi-agent behavioral models described by time-varying dynamical systems with neural networks. For efficient and interpretable learning, our model leverages theory-based architectures separating navigation and motion processes, and the theory-guided regularization for reliable behavioral modeling. This can provide interpretable signs of Granger-causal effects over time, i.e., when specific others cause the approach or separation. In experiments using synthetic datasets, our method achieved better performance than various baselines. We then analyzed multi-animal datasets of mice, flies, birds, and bats, which verified our method and obtained novel biological insights.

Denoising diffusion probabilistic models (DDPMs) (Ho et al. 2020) have shown impressive results on image and waveform generation in continuous state spaces. Here, we introduce Discrete Denoising Diffusion Probabilistic Models (D3PMs), diffusion-like generative models for discrete data that generalize the multinomial diffusion model of Hoogeboom et al. 2021, by going beyond corruption processes with uniform transition probabilities. This includes corruption with transition matrices that mimic Gaussian kernels in continuous space, matrices based on nearest neighbors in embedding space, and matrices that introduce absorbing states. The third allows us to draw a connection between diffusion models and autoregressive and mask-based generative models. We show that the choice of transition matrix is an important design decision that leads to improved results in image and text domains. We also introduce a new loss function that combines the variational lower bound with an auxiliary cross entropy loss. For text, this model class achieves strong results on character-level text generation while scaling to large vocabularies on LM1B. On the image dataset CIFAR-10, our models approach the sample quality and exceed the log-likelihood of the continuous-space DDPM model.

Abbreviation is Artificial Intelligence on a digital globe background. A power plant that will run on "artificial intelligence" is about to get underway in West Africa. The joint venture between Swiss-based Xcell Security House and Finance and U.S.-based Beyond Limits will embed intelligence and awareness into the operations -- something that will create more efficiencies, greater productivity, and increased environmental protections. When ordinary people hear about artificial intelligence -- AI for short -- they immediately think about how machines will replace humans. But as the experts explained to this reporter, AI is meant to eliminate "mundane activities" so that those running heavy industrial operations can solve problems and improve performance, which translates into healthier bottom lines.

Advanced cybersecurity capabilities are essential to safeguard software, systems, and data in a new era of cloud, the internet of things, and other smart technologies. In the real estate industry, for example, companies are concerned about the potential for hijacked elevators, as well as compromised building management and heating and cooling systems. According to Greg Belanger, vice president of security technologies at CBRE, the world's largest commercial real estate services and investment company, securing the enterprise has grown more complex--security teams must be familiar with controls and hardware on new devices, as well as what version of firmware is installed and what vulnerabilities are present. For example, if a heating, ventilation, and air-conditioning (HVAC) system is connected to the internet, he questions, "Is the firmware that's running the HVAC system vulnerable to attack? Could you find a way to traverse that network and come in and attack employees of that company?" Understanding enterprise vulnerabilities are crucial to safeguard physical assets but investing in the right tools can also be a challenge, says Belanger. "Artificial intelligence and machine learning need large sets of data to be effective in delivering the insights," he explains. In the era of cloud-first and industrial internet of things, the perimeter is becoming far more fluid. By applying AI and machine learning to data sets, he says, "You start to see patterns of risk and risky behavior start to emerge." Another priority when securing physical assets is to translate insights into metrics that C-suite leaders can understand, to help boost decision-making. CEOs and members of boards of directors, who are becoming more security savvy, can benefit from aggregated scores for attack surface management. "Everybody wants to know, especially after an attack like Colonial Pipeline, could that happen to us? How secure are we?" says Belanger.