An individualized treatment rule (ITR) is a decision rule that recommends treatments for patients based on their individual feature variables. In many practices, the ideal ITR for the primary outcome is also expected to cause minimal harm to other secondary outcomes. Therefore, our objective is to learn an ITR that not only maximizes the value function for the primary outcome, but also approximates the optimal rule for the secondary outcomes as closely as possible. To achieve this goal, we introduce a fusion penalty to encourage the ITRs based on different outcomes to yield similar recommendations. Two algorithms are proposed to estimate the ITR using surrogate loss functions. We prove that the agreement rate between the estimated ITR of the primary outcome and the optimal ITRs of the secondary outcomes converges to the true agreement rate faster than if the secondary outcomes are not taken into consideration. Furthermore, we derive the non-asymptotic properties of the value function and misclassification rate for the proposed method. Finally, simulation studies and a real data example are used to demonstrate the finite-sample performance of the proposed method.

We propose a novel semi-parametric classifier based on Mahalanobis distances of an observation from the competing classes. Our tool is a generalized additive model with the logistic link function that uses these distances as features to estimate the posterior probabilities of the different classes. While popular parametric classifiers like linear and quadratic discriminant analyses are mainly motivated by the normality of the underlying distributions, the proposed classifier is more flexible and free from such parametric assumptions. Since the densities of elliptic distributions are functions of Mahalanobis distances, this classifier works well when the competing classes are (nearly) elliptic. In such cases, it often outperforms popular nonparametric classifiers, especially when the sample size is small compared to the dimension of the data. To cope with non-elliptic and possibly multimodal distributions, we propose a local version of the Mahalanobis distance. Subsequently, we propose another classifier based on a generalized additive model that uses the local Mahalanobis distances as features. This nonparametric classifier usually performs like the Mahalanobis distance based semiparametric classifier when the underlying distributions are elliptic, but outperforms it for several non-elliptic and multimodal distributions. We also investigate the behaviour of these two classifiers in high dimension, low sample size situations. A thorough numerical study involving several simulated and real datasets demonstrate the usefulness of the proposed classifiers in comparison to many state-of-the-art methods.

Reconstruction attacks and defenses are essential in understanding the data leakage problem in machine learning. However, prior work has centered around empirical observations of gradient inversion attacks, lacks theoretical groundings, and was unable to disentangle the usefulness of defending methods versus the computational limitation of attacking methods. In this work, we propose a strong reconstruction attack in the setting of federated learning. The attack reconstructs intermediate features and nicely integrates with and outperforms most of the previous methods. On this stronger attack, we thoroughly investigate both theoretically and empirically the effect of the most common defense methods. Our findings suggest that among various defense mechanisms, such as gradient clipping, dropout, additive noise, local aggregation, etc., gradient pruning emerges as the most effective strategy to defend against state-of-the-art attacks.

We propose RanDumb to examine the efficacy of continual representation learning. RanDumb embeds raw pixels using a fixed random transform which approximates an RBF-Kernel, initialized before seeing any data, and learns a simple linear classifier on top. We present a surprising and consistent finding: RanDumb significantly outperforms the continually learned representations using deep networks across numerous continual learning benchmarks, demonstrating the poor performance of representation learning in these scenarios. RanDumb stores no exemplars and performs a single pass over the data, processing one sample at a time. It complements GDumb, operating in a low-exemplar regime where GDumb has especially poor performance. We reach the same consistent conclusions when RanDumb is extended to scenarios with pretrained models replacing the random transform with pretrained feature extractor. Our investigation is both surprising and alarming as it questions our understanding of how to effectively design and train models that require efficient continual representation learning, and necessitates a principled reinvestigation of the widely explored problem formulation itself. Our code is available at https://github.com/drimpossible/RanDumb.

Chest X-ray is a commonly used tool during triage, diagnosis and management of respiratory diseases. In resource-constricted settings, optimizing this resource can lead to valuable cost savings for the health care system and the patients as well as to and improvement in consult time. We used prospectively-collected data from 137 patients referred for chest X-ray at the Christian Medical Center and Hospital (CMCH) in Purnia, Bihar, India. Each patient provided at least five coughs while awaiting radiography. Collected cough sounds were analyzed using acoustic AI methods. Cross-validation was done on temporal and spectral features on the cough sounds of each patient. Features were summarized using standard statistical approaches. Three models were developed, tested and compared in their capacity to predict an abnormal result in the chest X-ray. All three methods yielded models that could discriminate to some extent between normal and abnormal with the logistic regression performing best with an area under the receiver operating characteristic curves ranging from 0.7 to 0.78. Despite limitations and its relatively small sample size, this study shows that AI-enabled algorithms can use cough sounds to predict which individuals presenting for chest radiographic examination will have a normal or abnormal results. These results call for expanding this research given the potential optimization of limited health care resources in low- and middle-income countries.

This paper reports our work on building up a Cantonese Speech-to-Text (STT) system with a syllable based acoustic model. This is a part of an effort in building a STT system to aid dyslexic students who have cognitive deficiency in writing skills but have no problem expressing their ideas through speech. For Cantonese speech recognition, the basic unit of acoustic models can either be the conventional Initial-Final (IF) syllables, or the Onset-Nucleus-Coda (ONC) syllables where finals are further split into nucleus and coda to reflect the intra-syllable variations in Cantonese. By using the Kaldi toolkit, our system is trained using the stochastic gradient descent optimization model with the aid of GPUs for the hybrid Deep Neural Network and Hidden Markov Model (DNN-HMM) with and without I-vector based speaker adaptive training technique. The input features of the same Gaussian Mixture Model with speaker adaptive training (GMM-SAT) to DNN are used in all cases. Experiments show that the ONC-based syllable acoustic modeling with I-vector based DNN-HMM achieves the best performance with the word error rate (WER) of 9.66% and the real time factor (RTF) of 1.38812.

The volume of image repositories continues to grow. Despite the availability of content-based addressing, we still lack a lightweight tool that allows us to discover images of distinct characteristics from a large collection. In this paper, we propose a fast and training-free algorithm for novel image discovery. The key of our algorithm is formulating a collection of images as a perceptual distance-weighted graph, within which our task is to locate the K-densest subgraph that corresponds to a subset of the most unique images. While solving this problem is not just NP-hard but also requires a full computation of the potentially huge distance matrix, we propose to relax it into a K-sparse eigenvector problem that we can efficiently solve using stochastic gradient descent (SGD) without explicitly computing the distance matrix. We compare our algorithm against state-of-the-arts on both synthetic and real datasets, showing that it is considerably faster to run with a smaller memory footprint while able to mine novel images more accurately.

Identifying how much a model ${\widehat{p}}_{\theta}(Y|X)$ knows about the stochastic real-world process $p(Y|X)$ it was trained on is important to ensure it avoids producing incorrect or "hallucinated" answers or taking unsafe actions. But this is difficult for generative models because probabilistic predictions do not distinguish between per-response noise (aleatoric uncertainty) and lack of knowledge about the process (epistemic uncertainty), and existing epistemic uncertainty quantification techniques tend to be overconfident when the model underfits. We propose a general strategy for teaching a model to both approximate $p(Y|X)$ and also estimate the remaining gaps between ${\widehat{p}}_{\theta}(Y|X)$ and $p(Y|X)$: train it to predict pairs of independent responses drawn from the true conditional distribution, allow it to "cheat" by observing one response while predicting the other, then measure how much it cheats. Remarkably, we prove that being good at cheating (i.e. cheating whenever it improves your prediction) is equivalent to being second-order calibrated, a principled extension of ordinary calibration that allows us to construct provably-correct frequentist confidence intervals for $p(Y|X)$ and detect incorrect responses with high probability. We demonstrate empirically that our approach accurately estimates how much models don't know across ambiguous image classification, (synthetic) language modeling, and partially-observable navigation tasks, outperforming existing techniques.

We study quantum neural networks made by parametric one-qubit gates and fixed two-qubit gates in the limit of infinite width, where the generated function is the expectation value of the sum of single-qubit observables over all the qubits. First, we prove that the probability distribution of the function generated by the untrained network with randomly initialized parameters converges in distribution to a Gaussian process whenever each measured qubit is correlated only with few other measured qubits. Then, we analytically characterize the training of the network via gradient descent with square loss on supervised learning problems. We prove that, as long as the network is not affected by barren plateaus, the trained network can perfectly fit the training set and that the probability distribution of the function generated after training still converges in distribution to a Gaussian process. Finally, we consider the statistical noise of the measurement at the output of the network and prove that a polynomial number of measurements is sufficient for all the previous results to hold and that the network can always be trained in polynomial time.

Statistical learning theory is traditionally founded on the assumption of a static data distribution, where statistical models are trained and deployed in the same environment. However, this assumption is often violated in practice, where the data distribution keeps changing over time. The temporal distribution shift can lead to serious decline in model performance post-deployment, which underlines the critical need to monitor models and detect potential degradation. Moreover, one often needs to choose among multiple candidate models originating from different learning algorithms (e.g., linear regression, random forests, neural networks) and hyperparameters (e.g., penalty parameter, step size, time window for training). Temporal distribution shift poses a major challenge to model selection, as past performance may not reliably predict future outcomes. Learners usually have to work with limited data from the current time period and abundant historical data, whose distributions may vary significantly.