Various logit-adjusted parameterizations of the cross-entropy (CE) loss have been proposed as alternatives to weighted CE for training large models on label-imbalanced data far beyond the zero train error regime. The driving force behind those designs has been the theory of implicit bias, which for linear(ized) models, explains why they successfully induce bias on the optimization path towards solutions that favor minorities. Aiming to extend this theory to non-linear models, we investigate the implicit geometry of classifiers and embeddings that are learned by different CE parameterizations. Our main result characterizes the global minimizers of a non-convex cost-sensitive SVM classifier for the unconstrained features model, which serves as an abstraction of deep nets. We derive closed-form formulas for the angles and norms of classifiers and embeddings as a function of the number of classes, the imbalance and the minority ratios, and the loss hyperparameters. Using these, we show that logit-adjusted parameterizations can be appropriately tuned to learn symmetric geometries irrespective of the imbalance ratio. We complement our analysis with experiments and an empirical study of convergence accuracy in deep-nets.

Heart failure is a serious and life-threatening condition that can lead to elevated pressure in the left ventricle. Pulmonary Arterial Wedge Pressure (PAWP) is an important surrogate marker indicating high pressure in the left ventricle. PAWP is determined by Right Heart Catheterization (RHC) but it is an invasive procedure. A non-invasive method is useful in quickly identifying high-risk patients from a large population. In this work, we develop a tensor learning-based pipeline for identifying PAWP from multimodal cardiac Magnetic Resonance Imaging (MRI). This pipeline extracts spatial and temporal features from high-dimensional scans. For quality control, we incorporate an epistemic uncertainty-based binning strategy to identify poor-quality training samples. To improve the performance, we learn complementary information by integrating features from multimodal data: cardiac MRI with short-axis and four-chamber views, and Electronic Health Records. The experimental analysis on a large cohort of $1346$ subjects who underwent the RHC procedure for PAWP estimation indicates that the proposed pipeline has a diagnostic value and can produce promising performance with significant improvement over the baseline in clinical practice (i.e., $\Delta$AUC $=0.10$, $\Delta$Accuracy $=0.06$, and $\Delta$MCC $=0.39$). The decision curve analysis further confirms the clinical utility of our method.

Label distribution learning (LDL) is an effective method to predict the label description degree (a.k.a. label distribution) of a sample. However, annotating label distribution (LD) for training samples is extremely costly. So recent studies often first use label enhancement (LE) to generate the estimated label distribution from the logical label and then apply external LDL algorithms on the recovered label distribution to predict the label distribution for unseen samples. But this step-wise manner overlooks the possible connections between LE and LDL. Moreover, the existing LE approaches may assign some description degrees to invalid labels. To solve the above problems, we propose a novel method to learn an LDL model directly from the logical label, which unifies LE and LDL into a joint model, and avoids the drawbacks of the previous LE methods. Extensive experiments on various datasets prove that the proposed approach can construct a reliable LDL model directly from the logical label, and produce more accurate label distribution than the state-of-the-art LE methods.

Many planning and decision activities in logistics and supply chain management are based on forecasts of multiple time dependent factors. Therefore, the quality of planning depends on the quality of the forecasts. We compare various forecasting methods in terms of out of the box forecasting performance on a broad set of simulated time series. We simulate various linear and non-linear time series and look at the one step forecast performance of statistical learning methods.

Achieving optimal statistical performance while ensuring the privacy of personal data is a challenging yet crucial objective in modern data analysis. However, characterizing the optimality, particularly the minimax lower bound, under privacy constraints is technically difficult. To address this issue, we propose a novel approach called the score attack, which provides a lower bound on the differential-privacy-constrained minimax risk of parameter estimation. The score attack method is based on the tracing attack concept in differential privacy and can be applied to any statistical model with a well-defined score statistic. It can optimally lower bound the minimax risk of estimating unknown model parameters, up to a logarithmic factor, while ensuring differential privacy for a range of statistical problems. We demonstrate the effectiveness and optimality of this general method in various examples, such as the generalized linear model in both classical and high-dimensional sparse settings, the Bradley-Terry-Luce model for pairwise comparisons, and nonparametric regression over the Sobolev class.

Modeling room acoustics in a field setting involves some degree of blind parameter estimation from noisy and reverberant audio. Modern approaches leverage convolutional neural networks (CNNs) in tandem with time-frequency representation. Using short-time Fourier transforms to develop these spectrogram-like features has shown promising results, but this method implicitly discards a significant amount of audio information in the phase domain. Inspired by recent works in speech enhancement, we propose utilizing novel phase-related features to extend recent approaches to blindly estimate the so-called "reverberation fingerprint" parameters, namely, volume and RT60. The addition of these features is shown to outperform existing methods that rely solely on magnitude-based spectral features across a wide range of acoustics spaces. We evaluate the effectiveness of the deployment of these novel features in both single-parameter and multi-parameter estimation strategies, using a novel dataset that consists of publicly available room impulse responses (RIRs), synthesized RIRs, and in-house measurements of real acoustic spaces.

The novel coronavirus (COVID-19) was declared as a global pandemic by the World Health Organization (WHO) after it was first discovered in Wuhan, China [1]. Over one year, the virus has infected more than 68 million people worldwide [2]. The virus can be fatal for elderly people or ones with chronic diseases [3]. Different countries across the globe have imposed several social practices and strategies to reduce the spread of the infection and to ensure the well-being of the residents. These practices and strategies include but are not limited to social distancing, restricted and authorized travels, remote work and education, reduced working staff in organizations, and frequent COVID-19 tests. These measures have been proved potential in reducing the disease spread and death in the previous pandemics [3], [4]. Several studies have focused on machine learning time series models to forecast the number of COVID-19 infections in different countries [5, 6, 7, 8, 9, 10, 11, 12, 13, 14]. This is to aid the government in designing and regulating efficient virus spread-mitigating strategies and to enable healthcare organizations for effective planning of health personnel and facilities resources. Based on the forecasted infections, the government can either make the confinement laws stricter or can ease them.

Lung cancer is the leading cause of cancer death in the world. Accurate determination of the EGFR (epidermal growth factor receptor) mutation status is highly relevant for the proper treatment of this patients. Purpose: The aim of this study was to predict the mutational status of the EGFR in non-small cell lung cancer patients using radiomics features extracted from PET-CT images. Methods: Retrospective study that involve 34 patients with lung cancer confirmed by histology and EGFR status mutation assessment. A total of 2.205 radiomics features were extracted from manual segmentation of the PET-CT images using pyradiomics library. Both computed tomography and positron emission tomography images were used. All images were acquired with intravenous iodinated contrast and F18-FDG. Preprocessing includes resampling, normalization, and discretization of the pixel intensity. Three methods were used for the feature selection process: backward selection (set 1), forward selection (set 2), and feature importance analysis of random forest model (set 3). Nine machine learning methods were used for radiomics model building. Results: 35.2% of patients had EGFR mutation, without significant differences in age, gender, tumor size and SUVmax. After the feature selection process 6, 7 and 17 radiomics features were selected, respectively in each group. The best performances were obtained by Ridge Regression in set 1: AUC of 0.826 (95% CI, 0.811 - 0.839), Random Forest in set 2: AUC of 0.823 (95% CI, 0.808 - 0.838) and Neural Network in set 3: AUC of 0.821 (95% CI, 0.808 - 0.835). Conclusion: The radiomics features analysis has the potential of predicting clinically relevant mutations in lung cancer patients through a non-invasive methodology.

Many practitioners in robotics regularly depend on classic, hand-designed algorithms. Often the performance of these algorithms is tuned across a dataset of annotated examples which represent typical deployment conditions. Automatic tuning of these settings is traditionally known as algorithm configuration. In this work, we extend algorithm configuration to automatically discover multiple modes in the tuning dataset. Unlike prior work, these configuration modes represent multiple dataset instances and are detected automatically during the course of optimization. We propose three methods for mode discovery: a post hoc method, a multi-stage method, and an online algorithm using a multi-armed bandit. Our results characterize these methods on synthetic test functions and in multiple robotics application domains: stereoscopic depth estimation, differentiable rendering, motion planning, and visual odometry. We show the clear benefits of detecting multiple modes in algorithm configuration space.

With high device integration density and evolving sophisticated device structures in semiconductor chips, detecting defects becomes elusive and complex. Conventionally, machine learning (ML)-guided failure analysis is performed with offline batch mode training. However, the occurrence of new types of failures or changes in the data distribution demands retraining the model. During the manufacturing process, detecting defects in a single-pass online fashion is more challenging and favoured. This paper focuses on novel quantile online learning for semiconductor failure analysis. The proposed method is applied to semiconductor device-level defects: FinFET bridge defect, GAA-FET bridge defect, GAA-FET dislocation defect, and a public database: SECOM. From the obtained results, we observed that the proposed method is able to perform better than the existing methods. Our proposed method achieved an overall accuracy of 86.66% and compared with the second-best existing method it improves 15.50% on the GAA-FET dislocation defect dataset.