Search for: All records

Abstract Digital biomarkers can radically transform the standard of care for chronic conditions that are complex to manage. In this work, we propose a scalable computational framework for discovering digital biomarkers of glycemic control. As a feasibility study, we leveraged over 79,000 days of digital data to define objective features, model the impact of each feature, classify glycemic control, and identify the most impactful digital biomarkers. Our research shows that glycemic control varies by age group, and was worse in the youngest population of subjects between the ages of 2–14. In addition, digital biomarkers like prior-day time above range and prior-day time in range, as well as total daily bolus and total daily basal were most predictive of impending glycemic control. With a combination of the top-ranked digital biomarkers, we achieved an average F1 score of 82.4% and 89.7% for classifying next-day glycemic control across two unique datasets. 

Lion (Evolved Sign Momentum), a new optimizer discovered through program search, has shown promising results in training large AI models. It performs comparably or favorably to AdamW but with greater memory efficiency. As we can expect from the results of a random search program, Lion incorporates elements from several existing algorithms, including signed momentum, decoupled weight decay, Polak, and Nesterov momentum, but does not fit into any existing category of theoretically grounded optimizers. Thus, even though Lion appears to perform well as a general-purpose optimizer for a wide range of tasks, its theoretical basis remains uncertain. This lack of theoretical clarity limits opportunities to further enhance and expand Lion's efficacy. This work aims to demystify Lion. Based on both continuous-time and discrete-time analysis, we demonstrate that Lion is a theoretically novel and principled approach for minimizing a general loss function $f(x)$ while enforcing a bound constraint $$\norm{x}_\infty \leq 1/\lambda$$. Lion achieves this through the incorporation of decoupled weight decay, where $$\lambda$$ represents the weight decay coefficient. Our analysis is made possible by the development of a new Lyapunov function for the Lion updates. It applies to a broader family of Lion-$$\phi$$ algorithms, where the $$\text{sign}(\cdot)$$ operator in Lion is replaced by the subgradient of a convex function $$\phi$$, leading to the solution of a general composite optimization problem of $$\min_x f(x) + \phi^*(x)$$. Our findings provide valuable insights into the dynamics of Lion and pave the way for further improvements and extensions of Lion-related algorithms. 

We present rectified flow, a surprisingly simple approach to learning (neural) ordinary differential equation (ODE) models to transport between two empirically observed distributions π0 and π1, hence providing a unified solution to generative modeling and domain transfer, among various other tasks involving distribution transport. The idea of rectified flow is to learn the ODE to follow the straight paths connecting the points drawn from π0 and π1 as much as possible. This is achieved by solving a straightforward nonlinear least squares optimization problem, which can be easily scaled to large models without introducing extra parameters beyond standard supervised learning. The straight paths are special and preferred because they are the shortest paths between two points, and can be simulated exactly without time discretization and hence yield computationally efficient models. We show that the procedure of learning a rectified flow from data, called rectification, turns an arbitrary coupling of π0 and π1 to a new deterministic coupling with provably non-increasing convex transport costs. In addition, recursively applying rectification allows us to obtain a sequence of flows with increasingly straight paths, which can be simulated accurately with coarse time discretization in the inference phase. In empirical studies, we show that rectified flow performs superbly on image generation, image-to-image translation, and domain adaptation. In particular, on image generation and translation, our method yields nearly straight flows that give high quality results even with a single Euler discretization step. 

Bilevel optimization (BO) is useful for solving a variety of important machine learning problems including but not limited to hyperparameter optimization, meta- learning, continual learning, and reinforcement learning. Conventional BO methods need to differentiate through the low-level optimization process with implicit dif- ferentiation, which requires expensive calculations related to the Hessian matrix. There has been a recent quest for first-order methods for BO, but the methods pro- posed to date tend to be complicated and impractical for large-scale deep learning applications. In this work, we propose a simple first-order BO algorithm that de- pends only on first-order gradient information, requires no implicit differentiation, and is practical and efficient for large-scale non-convex functions in deep learning. We provide a non-asymptotic convergence analysis of the proposed method to stationary points for non-convex objectives and present empirical results that show its superior practical performance. 

Face touch is an unconscious human habit. Frequent touching of sensitive/mucosal facial zones (eyes, nose, and mouth) increases health risks by passing pathogens into the body and spreading diseases. Furthermore, accurate monitoring of face touch is critical for behavioral intervention. Existing monitoring systems only capture objects approaching the face, rather than detecting actual touches. As such, these systems are prone to false positives upon hand or object movement in proximity to one's face (e.g., picking up a phone). We present FaceSense, an ear-worn system capable of identifying actual touches and differentiating them between sensitive/mucosal areas from other facial areas. Following a multimodal approach, FaceSense integrates low-resolution thermal images and physiological signals. Thermal sensors sense the thermal infrared signal emitted by an approaching hand, while physiological sensors monitor impedance changes caused by skin deformation during a touch. Processed thermal and physiological signals are fed into a deep learning model (TouchNet) to detect touches and identify the facial zone of the touch. We fabricated prototypes using off-the-shelf hardware and conducted experiments with 14 participants while they perform various daily activities (e.g., drinking, talking). Results show a macro-F1-score of 83.4% for touch detection with leave-one-user-out cross-validation and a macro-F1-score of 90.1% for touch zone identification with a personalized model. 

Data augmentation (DA) is an essential technique for training state-of-the-art deep learning systems. In this paper, we empirically show that the standard data augmentation methods may introduce distribution shift and consequently hurt the performance on unaugmented data during inference. To alleviate this issue, we propose a simple yet effective approach, dubbed KeepAugment, to increase the fidelity of augmented images. The idea is to use the saliency map to detect important regions on the original images and preserve these informative regions during augmentation. This information-preserving strategy allows us to generate more faithful training examples. Empirically, we demonstrate that our method significantly improves upon a number of prior art data augmentation schemes, e.g. AutoAugment, Cutout, random erasing, achieving promising results on image classification, semi-supervised image classification, multi-view multi-camera tracking and object detection. 

We propose MaxUp, an embarrassingly simple, highly effective technique for improving the generalization performance of machine learning models, especially deep neural networks. The idea is to generate a set of augmented data with some random perturbations or transforms and minimize the maximum, or worst case loss over the augmented data. By doing so, we implicitly introduce a smoothness or robustness regularization against the random perturbations, and hence improve the generation performance. For example, in the case of Gaussian perturbation, MaxUp is asymptotically equivalent to using the gradient norm of the loss as a penalty to encourage smoothness. We test MaxUp on a range of tasks, including image classification, language modeling, and adversarial certification, on which MaxUp consistently outperforms the existing best baseline methods, without introducing substantial computational overhead. In particular, we improve ImageNet classification from the state-of-the-art top-1 accuracy 85.5% without extra data to 85.8%. Code will be released soon. 

Semi-supervised learning (SSL) is a key approach toward more data-efficient machine learning by jointly leverage both labeled and unlabeled data. We propose AlphaMatch, an efficient SSL method that leverages data augmentations, by efficiently enforcing the label consistency between the data points and the augmented data derived from them. Our key technical contribution lies on: 1) using alpha-divergence to prioritize the regularization on data with high confidence, achieving a similar effect as FixMatch but in a more flexible fashion, and 2) proposing an optimization-based, EM-like algorithm to enforce the consistency, which enjoys better convergence than iterative regularization procedures used in recent SSL methods such as FixMatch, UDA, and MixMatch. AlphaMatch is simple and easy to implement, and consistently outperforms prior arts on standard benchmarks, e.g. CIFAR-10, SVHN, CIFAR-100, STL-10. Specifically, we achieve 91.3 data per class, substantially improving over the previously best 88.7 achieved by FixMatch.