Search for: All records

Abstract Analytic continuation from (3, 1) signature Minkowski to (2, 2) signature Klein space has emerged as a useful tool for the understanding of scattering amplitudes and flat space holography. Under this continuation, past and future null infinity merge into a single boundary ( $J$) which is the product of a null line with a (1, 1) signature torus. The Minkowskian $S$-matrix continues to a Kleinian $S$-vector which in turn may be represented by a Poincaré-invariant vacuum state $|C⟩$in the Hilbert space built on $J$. $|C⟩$contains all information about $S$in a novel, repackaged form. We give an explicit construction of $|C⟩$in a Lorentz/conformal basis for a free massless scalar. $J$separates into two halves $J_{\pm }$which are the asymptotic null boundaries of the regions timelike and spacelike separated from the origin. $|C⟩$is shown to be a maximally entangled state in the product of the $J_{\pm }$Hilbert spaces. 

A<sc>bstract</sc> We determine tree level, all-order celestial operator product expansions (OPEs) of gluons and gravitons in the maximally helicity violating (MHV) sector. We start by obtaining the all-order collinear expansions of MHV amplitudes using the inverse soft recursion relations that they satisfy. These collinear expansions are recast as celestial OPE expansions in bases of momentum as well as boost eigenstates. This shows that inverse soft recursion for MHV amplitudes is dual to OPE recursion in celestial conformal field theory. 

Federated learning—multi-party, distributed learning in a decentralized environment—is vulnerable to model poisoning attacks, more so than centralized learning. This is because malicious clients can collude and send in carefully tailored model updates to make the global model inaccurate. This motivated the development of Byzantine-resilient federated learning algorithms, such as Krum, Bulyan, FABA, and FoolsGold. However, a recently developed untargeted model poisoning attack showed that all prior defenses can be bypassed. The attack uses the intuition that simply by changing the sign of the gradient updates that the optimizer is computing, for a set of malicious clients, a model can be diverted from the optima to increase the test error rate. In this work, we develop FLAIR—a defense against this directed deviation attack (DDA), a state-of-the-art model poisoning attack. FLAIR is based on ourintuition that in federated learning, certain patterns of gradient flips are indicative of an attack. This intuition is remarkably stable across different learning algorithms, models, and datasets. FLAIR assigns reputation scores to the participating clients based on their behavior during the training phase and then takes a weighted contribution of the clients. We show that where the existing defense baselines of FABA [IJCAI’19], FoolsGold [Usenix ’20], and FLTrust [NDSS ’21] fail when 20-30% of the clients are malicious, FLAIR provides byzantine-robustness upto a malicious client percentage of 45%. We also show that FLAIR provides robustness against even a white-box version of DDA. 

Standard ML relies on training using a centrally collected dataset, while collaborative learning techniques such as Federated Learning (FL) enable data to remain decentralized at client locations. In FL, a central server coordinates the training process, reducing computation and communication expenses for clients. However, this centralization can lead to server congestion and heightened risk of malicious activity or data privacy breaches. In contrast, Peer-to-Peer Learning (P2PL) is a fully decentralized system where nodes manage both local training and aggregation tasks. While P2PL promotes privacy by eliminating the need to trust a single node, it also results in increased computation and communication costs, along with potential difficulties in achieving consensus among nodes. To address the limitations of both FL and P2PL, we propose a hybrid approach called Hubs-and-Spokes Learning (HSL). In HSL, hubs function similarly to FL servers, maintaining consensus but exerting less control over spokes. This paper argues that HSL’s design allows for greater availability and privacy than FL, while reducing computation and communication costs compared to P2PL. Additionally, HSL maintains consensus and integrity in the learning process. 

This study highlights innovative, minimally-invasive glucose sensing sutures for monitoring glucose levels in house sparrows. 

Abstract Chronic wounds present significant therapeutic challenges due to prolonged inflammation and bacterial infections, impeding healing. Conventional medicinal dressings typically deliver a single drug with a fixed release profile and lack responsiveness to variations in wound size, nature, or severity. This study introduces an innovative microneedle (MN) patch designed with different microneedle geometries and capable of dual‐drug delivery to address irregular wounds and complex therapeutic requirements. Utilizing CO₂ laser lithography, microneedle molds are fabricated with diverse geometries by precisely controlling laser parameters such as speed, power, and focus, achieving needle heights ranging from 162 ± 30 µm to 1570 ± 40 µm. The patch facilitates simultaneous delivery of simvastatin (SIM) for anti‐inflammatory and tetracycline hydrochloride (TH) for antibacterial properties, targeting different skin depths. In vitro diffusion studies confirm geometry‐dependent drug release profiles, with SIM achieving controlled release over three days and TH exhibiting sustained release over four days. Biocompatibility assays confirmed safety and enhanced fibroblast migration is noted in wound‐healing studies. Antimicrobial testing reveals a 99.9% reduction in bacterial viability. This cost‐effective and scalable approach enables precise, localized delivery and customization of MN arrays to match various wound geometries, offering a versatile platform for personalized medicine and improved chronic wound management.