Local news articles are a subset of news that impact users in a geographical area, such as a city, county, or state. Detecting local news (Step 1) and subsequently deciding its geographical location as well as radius of impact (Step 2) are two important steps towards accurate local news recommendation. Naive rule-based methods, such as detecting city names from the news title, tend to give erroneous results due to lack of understanding of the news content. Empowered by the latest development in natural language processing, we develop an integrated pipeline that enables automatic local news detection and content-based local news recommendations. In this paper, we focus on Step 1 of the pipeline, which highlights: (1) a weakly supervised framework incorporated with domain knowledge and auto data processing, and (2) scalability to multi-lingual settings. Compared with Stanford CoreNLP NER model, our pipeline has higher precision and recall evaluated on a real-world and human-labeled dataset. This pipeline has potential to more precise local news to users, helps local businesses get more exposure, and gives people more information about their neighborhood safety.

Machine learning models exhibit two seemingly contradictory phenomena: training data memorization, and various forms of forgetting. In memorization, models overfit specific training examples and become susceptible to privacy attacks. In forgetting, examples which appeared early in training are forgotten by the end. In this work, we connect these phenomena. We propose a technique to measure to what extent models "forget" the specifics of training examples, becoming less susceptible to privacy attacks on examples they have not seen recently. We show that, while non-convex models can memorize data forever in the worst-case, standard image, speech, and language models empirically do forget examples over time. We identify nondeterminism as a potential explanation, showing that deterministically trained models do not forget. Our results suggest that examples seen early when training with extremely large datasets - for instance those examples used to pre-train a model - may observe privacy benefits at the expense of examples seen later.

Recommender systems can strongly influence which information we see online, e.g., on social media, and thus impact our beliefs, decisions, and actions. At the same time, these systems can create substantial business value for different stakeholders. Given the growing potential impact of such AI-based systems on individuals, organizations, and society, questions of fairness have gained increased attention in recent years. However, research on fairness in recommender systems is still a developing area. In this survey, we first review the fundamental concepts and notions of fairness that were put forward in the area in the recent past. Afterward, through a review of more than 160 scholarly publications, we present an overview of how research in this field is currently operationalized, e.g., in terms of general research methodology, fairness measures, and algorithmic approaches. Overall, our analysis of recent works points to certain research gaps. In particular, we find that in many research works in computer science, very abstract problem operationalizations are prevalent and questions of the underlying normative claims and what represents a fair recommendation in the context of a given application are often not discussed in depth. These observations call for more interdisciplinary research to address fairness in recommendation in a more comprehensive and impactful manner.

Phone-level pronunciation scoring is a challenging task, with performance far from that of human annotators. Standard systems generate a score for each phone in a phrase using models trained for automatic speech recognition (ASR) with native data only. Better performance has been shown when using systems that are trained specifically for the task using non-native data. Yet, such systems face the challenge that datasets labelled for this task are scarce and usually small. In this paper, we present a transfer learning-based approach that leverages a model trained for ASR, adapting it for the task of pronunciation scoring. We analyze the effect of several design choices and compare the performance with a state-of-the-art goodness of pronunciation (GOP) system. Our final system is 20% better than the GOP system on EpaDB, a database for pronunciation scoring research, for a cost function that prioritizes low rates of unnecessary corrections.

A kriging-random forest hybrid model is developed for real-time ground property prediction ahead of the earth pressure balanced shield by integrating Kriging extrapolation and random forest, which can guide shield operating parameter selection thereby mitigate construction risks. The proposed KRF algorithm synergizes two types of information: prior information and real-time information. The previously predicted ground properties with EPB operating parameters are extrapolated via the Kriging algorithm to provide prior information for the prediction of currently being excavated ground properties. The real-time information refers to the real-time operating parameters of the EPB shield, which are input into random forest to provide a real-time prediction of ground properties. The integration of these two predictions is achieved by assigning weights to each prediction according to their uncertainties, ensuring the prediction of KRF with minimum uncertainty. The performance of the KRF algorithm is assessed via a case study of the Changsha Metro Line 4 project. It reveals that the proposed KRF algorithm can predict ground properties with an accuracy of 93%, overperforming the existing algorithms of LightGBM, AdaBoost-CART, and DNN by 29%, 8%, and 12%, respectively. Another dataset from Shenzhen Metro Line 13 project is utilized to further evaluate the model generalization performance, revealing that the model can transfer its learned knowledge from one region to another with an accuracy of 89%.

Car accidents remain a significant public safety issue worldwide, with the majority of them attributed to driver errors stemming from inadequate driving knowledge, non-compliance with regulations, and poor driving habits. To improve road safety, Driving Behavior Detection (DBD) systems have been proposed in several studies to identify safe and unsafe driving behavior. Many of these studies have utilized sensor data obtained from the Controller Area Network (CAN) bus to construct their models. However, the use of publicly available sensors is known to reduce the accuracy of detection models, while incorporating vendor-specific sensors into the dataset increases accuracy. To address the limitations of existing approaches, we present a reliable DBD system based on Graph Convolutional Long Short-Term Memory Networks (GConvLSTM) that enhances the precision and practicality of DBD models using public sensors. Additionally, we incorporate non-public sensors to evaluate the model's effectiveness. Our proposed model achieved a high accuracy of 97.5\% for public sensors and an average accuracy of 98.1\% for non-public sensors, indicating its consistency and accuracy in both settings. To enable local driver behavior analysis, we deployed our DBD system on a Raspberry Pi at the network edge, with drivers able to access daily driving condition reports, sensor data, and prediction results through a monitoring dashboard. Furthermore, the dashboard issues voice warnings to alert drivers of hazardous driving conditions. Our findings demonstrate that the proposed system can effectively detect hazardous and unsafe driving behavior, with potential applications in improving road safety and reducing the number of accidents caused by driver errors.

State-of-the-art natural language processing models have been shown to achieve remarkable performance in 'closed-world' settings where all the labels in the evaluation set are known at training time. However, in real-world settings, 'novel' instances that do not belong to any known class are often observed. This renders the ability to deal with novelties crucial. To initiate a systematic research in this important area of 'dealing with novelties', we introduce 'NoveltyTask', a multi-stage task to evaluate a system's performance on pipelined novelty 'detection' and 'accommodation' tasks. We provide mathematical formulation of NoveltyTask and instantiate it with the authorship attribution task that pertains to identifying the correct author of a given text. We use Amazon reviews corpus and compile a large dataset (consisting of 250k instances across 200 authors/labels) for NoveltyTask. We conduct comprehensive experiments and explore several baseline methods for the task. Our results show that the methods achieve considerably low performance making the task challenging and leaving sufficient room for improvement. Finally, we believe our work will encourage research in this underexplored area of dealing with novelties, an important step en route to developing robust systems.

State-of-the-art sequence-to-sequence models often require autoregressive decoding, which can be highly expensive. However, for some downstream tasks such as out-of-distribution (OOD) detection and resource allocation, the actual decoding output is not needed just a scalar attribute of this sequence. In these scenarios, where for example knowing the quality of a system's output to predict poor performance prevails over knowing the output itself, is it possible to bypass the autoregressive decoding? We propose Non-Autoregressive Proxy (NAP) models that can efficiently predict general scalar-valued sequence-level attributes. Importantly, NAPs predict these metrics directly from the encodings, avoiding the expensive autoregressive decoding stage. We consider two sequence-to-sequence task: Machine Translation (MT); and Automatic Speech Recognition (ASR). In OOD for MT, NAPs outperform a deep ensemble while being significantly faster. NAPs are also shown to be able to predict performance metrics such as BERTScore (MT) or word error rate (ASR). For downstream tasks, such as data filtering and resource optimization, NAPs generate performance predictions that outperform predictive uncertainty while being highly inference efficient.

Data Mining and Knowledge Discovery manuscript No. (will be inserted by the editor) Is AUC the best measure for practical comparison of anomaly detectors? Abstract The area under receiver operating characteristics (AUC) is the standard measure for comparison of anomaly detectors. Its advantage is in providing a scalar number that allows a natural ordering and is independent on a threshold, which allows to postpone the choice. In this work, we question whether AUC is a good metric for anomaly detection, or if it gives a false sense of comfort, due to relying on assumptions which are unlikely to hold in practice. Our investigation shows that variations of AUC emphasizing accuracy at low false positive rate seem to be better correlated with the needs of practitioners, but also that we can compare anomaly detectors only in the case when we have representative examples of anomalous samples. This last result is disturbing, as it suggests that in many cases, we should do active or few-show learning instead of pure anomaly detection. While this definition makes sense from the point of view of probability theory, the practitioners are interested only in anomalies of a certain kind. For example, the use of anomaly detection in computer security is motivated by the assumption that attacks are rare, but not every rare event is an attack. Anomaly detection has a long history and there are hundreds of models based on vastly different approaches, such as modifications of the k-nearest neighbors algorithm (Harmeling et al., 2006), random forests Figure 1 Receiver operating characteristic (ROC) curve and the corresponding AUC of a degenerate anomaly detector. FPR / TPR stands for false / true positive rate. Several comparative studies exist - e.g.

LASSO regularization is a popular regression tool to enhance the prediction accuracy of statistical models by performing variable selection through the $\ell_1$ penalty, initially formulated for the linear model and its variants. In this paper, the territory of LASSO is extended to two-layer ReLU neural networks, a fashionable and powerful nonlinear regression model. Specifically, given a neural network whose output $y$ depends only on a small subset of input $\boldsymbol{x}$, denoted by $\mathcal{S}^{\star}$, we prove that the LASSO estimator can stably reconstruct the neural network and identify $\mathcal{S}^{\star}$ when the number of samples scales logarithmically with the input dimension. This challenging regime has been well understood for linear models while barely studied for neural networks. Our theory lies in an extended Restricted Isometry Property (RIP)-based analysis framework for two-layer ReLU neural networks, which may be of independent interest to other LASSO or neural network settings. Based on the result, we advocate a neural network-based variable selection method. Experiments on simulated and real-world datasets show promising performance of the variable selection approach compared with existing techniques.