Search

This paper advances the theory and practice of Domain Generalization (DG) in machine learning. We consider the typical DG setting where the hypothesis is composed of a representation mapping followed by a labeling function. Within this setting, the majority of popular DG methods aim to jointly learn the representation and the labeling functions by minimizing a well-known upper bound for the classification risk in the unseen domain. In practice, however, methods based on this theoretical upper bound ignore a term that cannot be directly optimized due to its dual dependence on both the representation mapping and the unknown optimal labeling function in the unseen domain. To bridge this gap between theory and practice, we introduce a new upper bound that is free of terms having such dual dependence, resulting in a fully optimizable risk upper bound for the unseen domain. Our derivation leverages classical and recent transport inequalities that link optimal transport metrics with information-theoretic measures. Compared to previous bounds, our bound introduces two new terms: (i) the Wasserstein-2 barycenter term that aligns distributions between domains, and (ii) the reconstruction loss term that assesses the quality of representation in reconstructing the original data. Based on this new upper bound, we propose a novel DG algorithm named Wasserstein Barycenter Auto-Encoder (WBAE) that simultaneously minimizes the classification loss, the barycenter loss, and the reconstruction loss. Numerical results demonstrate that the proposed method outperforms current state-of-the-art DG algorithms on several datasets., This article has been accepted for publication in IEEE Access

Graphical abstract Diabetes mellitus is a metabolic disease manifested by hyperglycemia. For patients with type 1 and advanced type 2 diabetes mellitus, insulin therapy is essential. Subcutaneous injection remains the most common administration method. Non-invasive insulin delivery technologies are pursued because of their benefits of decreasing patients’ pain, anxiety, and stress. Transdermal delivery systems have gained extensive attention due to the ease of administration and absence of hepatic first-pass metabolism. Microneedle (MN) technology is one of the most promising tactics, which can effectively deliver insulin through skin stratum corneum in a minimally invasive and painless way. This article will review the research progress of MNs in insulin transdermal delivery, including hollow MNs, dissolving MNs, hydrogel MNs, and glucose-responsive MN patches, in which insulin dosage can be strictly controlled. The clinical studies about insulin delivery with MN devices have also been summarized and grouped based on the study phase. There are still several challenges to achieve successful translation of MNs-based insulin therapy. In this review, we also discussed these challenges including safety, efficacy, patient/prescriber acceptability, manufacturing and scale-up, and regulatory authority acceptability.

Invariance-principle-based methods such as Invariant Risk Minimization (IRM), have recently emerged as promising approaches for Domain Generalization (DG). Despite promising theory, such approaches fail in common classification tasks due to the mixing of true invariant features and spurious invariant features. To address this, we propose a framework based on the conditional entropy minimization (CEM) principle to filter-out the spurious invariant features leading to a new algorithm with a better generalization capability. We show that our proposed approach is closely related to the well-known Information Bottleneck (IB) framework and prove that under certain assumptions, entropy minimization can exactly recover the true invariant features. Our approach provides competitive classification accuracy compared to recent theoretically-principled state-of-the-art alternatives across several DG datasets., Comment: 10 pages, this paper was accepted at 26th International Conference on Pattern Recognition (ICPR-2022)

Common neutron detectors are sensitive to neutrons and γ-rays, especially the organic scintillation detectors. Therefore, it is necessary to separate the two signals in neutron detection. In this paper, five algorithms by discriminating the pulse shape were implemented and compared, including charge comparison algorithm, risetime algorithm, frequency gradient analysis algorithm, K-means++ clustering algorithm and the feature extraction method of BP neural networks algorithm was optimized. The results show that the risetime algorithm is the best in terms of FOM, and the processing time based on BP neural network is the fastest, and the DER of BP neural network algorithm is most small. Based on the above work, this paper can provide a basis for discrimination and optimization of n/γ discrimination algorithms in practical work.

Sensor platforms that exploit the fibrous textile threads as substrates offer great promise since they can be directly sewn, woven or stitched on to any clothing. They can also be placed directly in intimate contact with the skin. In this work, we present a thread-based sensing platform in the form of a multiplexed sensing patch for continuous simultaneous on-skin monitoring of sweat. The patch performs real-time, on-body measurements of important biomarkers present in sweat such as electrolytes (sodium and ammonium ions), metabolites (lactate) and acidity (pH). Flexible threads coated with conductive inks were used as sensing electrodes. Selective potentiometric detection of electrolytes and pH was made possible through ion-selective membrane deposition and pH-sensitive polyaniline coating on threads, respectively. An amperometric enzymatic sensing scheme with lactate oxidase was used for the detection of lactate. An array of the thread sensors is integrated onto a patch with connectivity to a miniaturized circuit module containing a potentiostat, microprocessor and wireless circuitry for wireless smartphone readout. Extensive in vitro validation and an in vivo human pilot study involving a maximal exertion test show the promise of this platform for real-time physiological monitoring of human performance/fitness under stress, as well as diagnostic monitoring through sweat analysis.

Threads have recently been demonstrated as thin and flexible substrates for diagnostics and wearable devices. These substrates provide flexible 3D integration with cloth using standard textile methods. Our group has recently demonstrated a pH mapping smart-bandage using thread-based polyaniline pH sensors for chronic wound monitoring. Here, we utilize high-surface area sensing regions to present an improved pH mapping bandage with more intimate wound contact and a faster response time compared to our previous work. Additionally, we condense our readout electronics into a 3D printed 25mm button form factor capable of easy integration with our thread-based sensors. We demonstrate our modified bandage's improved performance through sensitivity and response time measurements. Furthermore, we present pH measurements in spiked biological serum demonstrating that our bandage remains sensitive in complex biological environments.