Access to informative databases is a crucial part of notable research developments. In the field of domestic audio classification, there have been significant advances in recent years. Although several audio databases exist, these can be limited in terms of the amount of information they provide, such as the exact location of the sound sources, and the associated noise levels. In this work, we detail our approach on generating an unbiased synthetic domestic audio database, consisting of sound scenes and events, emulated in both quiet and noisy environments. Data is carefully curated such that it reflects issues commonly faced in a dementia patients environment, and recreate scenarios that could occur in real-world settings. Similarly, the room impulse response generated is based on a typical one-bedroom apartment at Hebrew SeniorLife Facility. As a result, we present an 11-class database containing excerpts of clean and noisy signals at 5-seconds duration each, uniformly sampled at 16 kHz. Using our baseline model using Continues Wavelet Transform Scalograms and AlexNet, this yielded a weighted F1-score of 86.24 percent.

This paper presents a new network intrusion detection system (NIDS) based on Graph Neural Networks (GNNs). GNNs are a relatively new sub-field of deep neural networks, which have the unique ability to leverage the inherent structure of graph-based data. Training and evaluation data for NIDSs are typically represented as flow records, which can naturally be represented in a graph format. This establishes the potential and motivation for exploring GNNs for the purpose of network intrusion detection, which is the focus of this paper. E-GraphSAGE, our proposed new approach is based on the established GraphSAGE model, but provides the necessary modifications in order to support edge features for edge classification, and hence the classification of network flows into benign and attack classes. An extensive experimental evaluation based on six recent NIDS benchmark datasets shows the excellent performance of our E-GraphSAGE based NIDS in comparison with the state-of-the-art.

Ever wondered what happens to a selfie you upload on a social media site? Activists and researchers have long warned about data privacy and said that photographs uploaded on the Internet may be used to train artificial intelligence (AI) powered facial recognition tools. These AI-enabled tools (such as Clearview, AWS Rekognition, Microsoft Azure, and Face) could in turn be used by governments or other institutions to track people and even draw conclusions such as the subject's religious or political preferences. Researchers have come up with ways to dupe or spoof these AI tools from being able to recognise or even detect a selfie, using adversarial attacks – or a way to alter input data that causes a deep-learning model to make mistakes. Two of these methods were presented last week at the International Conference of Learning Representations (ICLR), a leading AI conference that was held virtually.

Fractal's Computer Vision solution for Underwriting & Claims selected as one of the top solutions to be presented at the event to be attended by more than 400 P&C Execs Fractal's Chief Practice Officer for Insurance and Technology, Sankar Narayanan, will be speaking at "Tech Showcase: Computer Vision for Underwriting & Claims". The event will bring together over 400 top executives and insiders for discussions and presentations on some of the most pressing needs and innovative solutions. The talk will feature Fractal's Image and Video Analytics (IVA) solution and demonstrate how it enables #betterdecisions for P&C Insurers. IVA is Fractal's deep-learning based solution that is used by insurers to analyze images/videos/complex documents at scale, during underwriting and claims stages. During underwriting, it is used to assess value and estimate coverage with precision and during claims it is used to assess severity and estimate settlement amount, at speed.

The relevance of the Key Information Extraction (KIE) task is increasingly important in natural language processing problems. But there are still only a few well-defined problems that serve as benchmarks for solutions in this area. To bridge this gap, we introduce two new datasets (Kleister NDA and Kleister Charity). They involve a mix of scanned and born-digital long formal English-language documents. In these datasets, an NLP system is expected to find or infer various types of entities by employing both textual and structural layout features. The Kleister Charity dataset consists of 2,788 annual financial reports of charity organizations, with 61,643 unique pages and 21,612 entities to extract. The Kleister NDA dataset has 540 Non-disclosure Agreements, with 3,229 unique pages and 2,160 entities to extract. We provide several state-of-the-art baseline systems from the KIE domain (Flair, BERT, RoBERTa, LayoutLM, LAMBERT), which show that our datasets pose a strong challenge to existing models. The best model achieved an 81.77% and an 83.57% F1-score on respectively the Kleister NDA and the Kleister Charity datasets. We share the datasets to encourage progress on more in-depth and complex information extraction tasks.

The development of respiratory failure is common among patients in intensive care units (ICU). Large data quantities from ICU patient monitoring systems make timely and comprehensive analysis by clinicians difficult but are ideal for automatic processing by machine learning algorithms. Early prediction of respiratory system failure could alert clinicians to patients at risk of respiratory failure and allow for early patient reassessment and treatment adjustment. We propose an early warning system that predicts moderate/severe respiratory failure up to 8 hours in advance. Our system was trained on HiRID-II, a data-set containing more than 60,000 admissions to a tertiary care ICU. An alarm is typically triggered several hours before the beginning of respiratory failure. Our system outperforms a clinical baseline mimicking traditional clinical decision-making based on pulse-oximetric oxygen saturation and the fraction of inspired oxygen. To provide model introspection and diagnostics, we developed an easy-to-use web browser-based system to explore model input data and predictions visually.

Recent work has unveiled a theory for reasoning about the decisions made by binary classifiers: a classifier describes a Boolean function, and the reasons behind an instance being classified as positive are the prime-implicants of the function that are satisfied by the instance. One drawback of these works is that they do not explicitly treat scenarios where the underlying data is known to be constrained, e.g., certain combinations of features may not exist, may not be observable, or may be required to be disregarded. We propose a more general theory, also based on prime-implicants, tailored to taking constraints into account. The main idea is to view classifiers in the presence of constraints as describing partial Boolean functions, i.e., that are undefined on instances that do not satisfy the constraints. We prove that this simple idea results in reasons that are no less (and sometimes more) succinct. That is, not taking constraints into account (e.g., ignored, or taken as negative instances) results in reasons that are subsumed by reasons that do take constraints into account. We illustrate this improved parsimony on synthetic classifiers and classifiers learned from real data.

Abstract-- Applications that need to sense, measure, and gather real-time information from the environment frequently face three main restrictions: power consumption, cost, and lack of infrastructure. Most of the challenges imposed by these limitations can be better addressed by embedding Machine Learning (ML) classifiers in the hardware that senses the environment, creating smart sensors able to interpret the low-level data stream. However, for this approach to be cost-effective, we need highly efficient classifiers suitable to execute in unresourceful hardware, such as low-power microcontrollers. In this paper, we present an open-source tool named EmbML - Embedded Machine Learning that implements a pipeline to develop classifiers for resource-constrained hardware. We describe its implementation details and provide a comprehensive analysis of its classifiers considering accuracy, classification time, and memory usage. Moreover, we compare the performance of its classifiers with classifiers produced by related tools to demonstrate that our tool provides a diverse set of classification algorithms that are both compact and accurate. Therefore, these smart sensors are more powerefficient since they eliminate the need for communicating all the raw data. PPLICATIONS that need to sense, measure, and gather real-time information from the environment frequently of interest - e.g., a dry soil crop area that needs watering or face three main restrictions [1]: power consumption, cost, the capture of a disease-vector mosquito.

We present a novel method of performing spelling correction on short input strings, such as search queries or individual words. At its core lies a procedure for generating artificial typos which closely follow the error patterns manifested by humans. This procedure is used to train the production spelling correction model based on a transformer architecture. This model is currently served in the HubSpot product search. We show that our approach to typo generation is superior to the widespread practice of adding noise, which ignores human patterns. We also demonstrate how our approach may be extended to resource-scarce settings and train spelling correction models for Arabic, Greek, Russian, and Setswana languages, without using any labeled data.

In past years model-agnostic meta-learning (MAML) has been one of the most promising approaches in meta-learning. It can be applied to different kinds of problems, e.g., reinforcement learning, but also shows good results on few-shot learning tasks. Besides their tremendous success in these tasks, it has still not been fully revealed yet, why it works so well. Recent work proposes that MAML rather reuses features than rapidly learns. In this paper, we want to inspire a deeper understanding of this question by analyzing MAML's representation. We apply representation similarity analysis (RSA), a well-established method in neuroscience, to the few-shot learning instantiation of MAML. Although some part of our analysis supports their general results that feature reuse is predominant, we also reveal arguments against their conclusion. The similarity-increase of layers closer to the input layers arises from the learning task itself and not from the model. In addition, the representations after inner gradient steps make a broader change to the representation than the changes during meta-training.