Hit song prediction, one of the emerging fields in music information retrieval (MIR), remains a considerable challenge. Being able to understand what makes a given song a hit is clearly beneficial to the whole music industry. Previous approaches to hit song prediction have focused on using audio features of a record. This study aims to improve the prediction result of the top 10 hits among Billboard Hot 100 songs using more alternative metadata, including song audio features provided by Spotify, song lyrics, and novel metadata-based features (title topic, popularity continuity and genre class). Five machine learning approaches are applied, including: k-nearest neighbours, Naive Bayes, Random Forest, Logistic Regression and Multilayer Perceptron. Our results show that Random Forest (RF) and Logistic Regression (LR) with all features (including novel features, song audio features and lyrics features) outperforms other models, achieving 89.1% and 87.2% accuracy, and 0.91 and 0.93 AUC, respectively. Our findings also demonstrate the utility of our novel music metadata features, which contributed most to the models' discriminative performance.

Localization is a key challenge in many robotics applications. In this work we explore LIDAR-based global localization in both urban and natural environments and develop a method suitable for online application. Our approach leverages efficient deep learning architecture capable of learning compact point cloud descriptors directly from 3D data. The method uses an efficient feature space representation of a set of segmented point clouds to match between the current scene and the prior map. We show that down-sampling in the inner layers of the network can significantly reduce computation time without sacrificing performance. We present substantial evaluation of LIDAR-based global localization methods on nine scenarios from six datasets varying between urban, park, forest, and industrial environments. Part of which includes post-processed data from 30 sequences of the Oxford RobotCar dataset, which we make publicly available. Our experiments demonstrate a factor of three reduction of computation, 70% lower memory consumption with marginal loss in localization frequency. The proposed method allows the full pipeline to run on robots with limited computation payload such as drones, quadrupeds, and UGVs as it does not require a GPU at run time.

Current self-training methods such as standard self-training, co-training, tri-training, and others often focus on improving model performance on a single task, utilizing differences in input features, model architectures, and training processes. However, many tasks in natural language processing are about different but related aspects of language, and models trained for one task can be great teachers for other related tasks. In this work, we propose friend-training, a cross-task self-training framework, where models trained to do different tasks are used in an iterative training, pseudo-labeling, and retraining process to help each other for better selection of pseudo-labels. With two dialogue understanding tasks, conversational semantic role labeling and dialogue rewriting, chosen for a case study, we show that the models trained with the friend-training framework achieve the best performance compared to strong baselines.

Large language models, often trained with billions of parameters, have achieved impressive performance in recent years (Raffel et al., 2019) and are used in a wide variety of natural language generation tasks. However, their output is sometimes undesirable, with hallucinated content (Maynez et al., 2020; Filippova, 2020), and much work remains to fully understand their properties. In many applications, such as healthcare, question-answering systems, or customer service, incorrect predictions are particularly costly and must be avoided. This motivates the design of algorithms for large language models and other NLP tasks that achieve high precision on a large fraction of the input set, while abstaining on the rest. How can we devise such accurate models that allow a reject option?

Semi-supervised Learning (SSL) has witnessed great success owing to the impressive performances brought by various methods based on pseudo labeling and consistency regularization. However, we argue that existing methods might fail to utilize the unlabeled data more effectively since they either use a pre-defined / fixed threshold or an ad-hoc threshold adjusting scheme, resulting in inferior performance and slow convergence. We first analyze a motivating example to obtain intuitions on the relationship between the desirable threshold and model's learning status. Based on the analysis, we hence propose FreeMatch to adjust the confidence threshold in a self-adaptive manner according to the model's learning status. We further introduce a self-adaptive class fairness regularization penalty to encourage the model for diverse predictions during the early training stage. Extensive experiments indicate the superiority of FreeMatch especially when the labeled data are extremely rare. FreeMatch achieves 5.78%, 13.59%, and 1.28% error rate reduction over the latest state-of-the-art method FlexMatch on CIFAR-10 with 1 label per class, STL-10 with 4 labels per class, and ImageNet with 100 labels per class, respectively. Moreover, FreeMatch can also boost the performance of imbalanced SSL. The codes can be found at https://github.com/microsoft/Semi-supervised-learning.

Cerebral microbleeds detection is an important and challenging task. With the gaining popularity of the MRI, the ability to detect cerebral microbleeds also raises. Unfortunately, for radiologists, it is a time-consuming and laborious procedure. For this reason, various solutions to automate this process have been proposed for several years, but none of them is currently used in medical practice. In this context, the need to systematize the existing knowledge and best practices has been recognized as a factor facilitating the imminent synthesis of a real CMBs detection system practically applicable in medicine. To the best of our knowledge, all available publications regarding automatic cerebral microbleeds detection have been gathered, described, and assessed in this paper in order to distinguish the current research state and provide a starting point for future studies.

Falls have become more frequent in recent years, which has been harmful for senior citizens.Therefore detecting falls have become important and several data sets and machine learning model have been introduced related to fall detection. In this project report, a human fall detection method is proposed using a multi modality approach. We used the UP-FALL detection data set which is collected by dozens of volunteers using different sensors and two cameras. We use wrist sensor with acclerometer data keeping labels to binary classification, namely fall and no fall from the data set.We used fusion of camera and sensor data to increase performance. The experimental results shows that using only wrist data as compared to multi sensor for binary classification did not impact the model prediction performance for fall detection.

This paper studies a parametric family of algorithmic fairness metrics, called generalized entropy, which originally has been used in public welfare and recently introduced to machine learning community. As a meaningful metric to evaluate algorithmic fairness, it requires that generalized entropy specify fairness requirements of a classification problem and the fairness requirements should be realized with small deviation by an algorithm. We investigate the role of generalized entropy as a design parameter for fair classification algorithm through a fair empirical risk minimization with a constraint specified in terms of generalized entropy. We theoretically and experimentally study learnability of the problem.

Objective: Reliable tools to predict moyamoya disease (MMD) patients at risk for hemorrhage could have significant value. The aim of this paper is to develop three machine learning classification algorithms to predict hemorrhage in moyamoya disease. Methods: Clinical data of consecutive MMD patients who were admitted to our hospital between 2009 and 2015 were reviewed. Demographics, clinical, radiographic data were analyzed to develop artificial neural network (ANN), support vector machine (SVM), and random forest models. Results: We extracted 33 parameters, including 11 demographic and 22 radiographic features as input for model development. Of all compared classification results, ANN achieved the highest overall accuracy of 75.7% (95% CI, 68.6%-82.8%), followed by SVM with 69.2% (95% CI, 56.9%-81.5%) and random forest with 70.0% (95% CI, 57.0%-83.0%). Conclusions: The proposed ANN framework can be a potential effective tool to predict the possibility of hemorrhage among adult MMD patients based on clinical information and radiographic features.

The extensive damage caused by malware requires anti-malware systems to be constantly improved to prevent new threats. The current trend in malware detection is to employ machine learning models to aid in the classification process. We propose a new dataset with the objective of improving current anti-malware systems. The focus of this dataset is to improve host based intrusion detection systems by providing API call sequences for thousands of malware samples executed in Windows 10 virtual machines. A tutorial on how to create and expand this dataset is provided along with a benchmark demonstrating how to use this dataset to classify malware. The data contains long sequences of API calls for each sample, and in order to create models that can be deployed in resource constrained devices, three feature selection methods were tested. The principal innovation, however, lies in the multi-label classification system in which one sequence of APIs can be tagged with multiple labels describing its malicious behaviours.