Search for: All records

We provide new connections between two distinct federated learning approaches based on (i) ADMM and (ii) Variational Bayes (VB), and propose new variants by combining their complementary strengths. Specifically, we show that the dual variables in ADMM naturally emerge through the "site" parameters used in VB with isotropic Gaussian covariances. Using this, we derive two versions of ADMM from VB that use flexible covariances and functional regularisation, respectively. Through numerical experiments, we validate the improvements obtained in performance. The work shows connection between two fields that are believed to be fundamentally different and combines them to improve federated learning. 

Gradual typing has emerged as a promising typing discipline for reconciling static and dynamic typing, which have respective strengths and shortcomings. Thanks to its promises, gradual typing has gained tremendous momentum in both industry and academia. A main challenge in gradual typing is that, however, the performance of its programs can often be unpredictable, and adding or removing the type of a a single parameter may lead to wild performance swings. Many approaches have been proposed to optimize gradual typing performance, but little work has been done to aid the understanding of the performance landscape of gradual typing and navigating the migration process (which adds type annotations to make programs more static) to avert performance slowdowns. Motivated by this situation, this work develops a machine-learning-based approach to predict the performance of each possible way of adding type annotations to a program. On top of that, many supports for program migrations could be developed, such as finding the most performant neighbor of any given configuration. Our approach gauges runtime overheads of dynamic type checks inserted by gradual typing and uses that information to train a machine learning model, which is used to predict the running time of gradual programs. We have evaluated our approach on 12 Python benchmarks for both guarded and transient semantics. For guarded semantics, our evaluation results indicate that with only 40 training instances generated from each benchmark, the predicted times for all other instances differ on average by 4% from the measured times. For transient semantics, the time difference ratio is higher but the time difference is often within 0.1 seconds. 

Hierarchical surfaces comprised of both microscale and nanoscale structures have been previously studied as a means of targeting multiple length scales to achieve superior pool boiling performance. However, preceding studies have focused almost exclusively on high surface tension working fluids while technologically important low surface tension fluids have remained largely unexplored. In this work, we utilize scalable manufacturing techniques to realize four separate surface types (planar, nanoscale-modified, microscale-modified, and hierarchical) and experimentally determine their respective pool boiling performance within the low surface tension commercial working fluid HFE-7100. A maximum heat transfer enhancement of 125 % at 38 K of superheat was observed for the best performing samples, which interestingly were nanoscale-modified and not those of the hierarchical type. Visual observations via high-speed video analysis of vapor bubble behaviour are utilized to explain the underlying multiphase physics as to why these samples performed so well and future directions for achieving surface optimization across multiple length scales. 

This paper investigates how trust towards service providers and the adoption of privacy controls belonging to two specific purposes (control over “sharing” vs. “usage” of data) vary based on users’ technical literacy. Towards that, we chose Google as the context and conducted an online survey across 209 Google users. Our results suggest that integrity and benevolence perceptions toward Google are significantly lower among technical participants than non-technical participants. While trust perceptions differ between non-technical adopters and non-adopters of privacy controls, no such difference is found among the technical counterparts. Notably, among the non-technical participants, the direction of trust affecting privacy control adoption is observed to be reversed based on the purpose of the controls. Using qualitative analysis, we extract trust-enhancing and dampening factors contributing to users’ trusting beliefs towards Google’s protection of user privacy. The implications of our findings for the design and promotion of privacy controls are discussed in the paper. 

Riemannian submanifold optimization with momentum is computationally challenging because, to ensure that the iterates remain on the submanifold, we often need to solve difficult differential equations. Here, we simplify such difficulties for a class of structured symmetric positive-definite matrices with the affine-invariant metric. We do so by proposing a generalized version of the Riemannian normal coordinates that dynamically orthonormalizes the metric and locally converts the problem into an unconstrained problem in the Euclidean space. We use our approach to simplify existing approaches for structured covariances and develop matrix-inverse-free 2nd-order optimizers for deep learning in low precision settings. 

Hardware Trojans in Integrated Circuits (ICs), that are inserted as hostile modifications in the design phase and/or the fabrication phase, are a security threat since the semiconductor manufacturing process is increasingly becoming globalized. These Trojans are devised to stay hidden during standard structural and functional testing procedures and only activate under pre-determined rare conditions (e.g., after a large number of clock cycles or the assertion of an improbable net). Once triggered, they can deliver malicious payloads (e.g., denial-of-service and information leakage attacks). Current literature identifies a collection of logic Trojans (both trigger circuits and payloads), but minimal research exists on memory Trojans despite their high feasibility. Emerging Non-Volatile Memories (NVMs), such as Resistive RAM (RRAM), have special properties such as non-volatility and gradual drift in bitcell resistance under a pulsing voltage input that make them prime targets to deploy hardware Trojans. In this paper, we present two delay-based and two voltage-based Trojan triggers using emerging NVM (ENTT) by utilizing RRAM’s resistance drift under a pulsing voltage input. Simulations show that ENTTs can be triggered by reading/writing to a specific memory address N times (N could be 2,500–3,500 or a different value for each ENTT design). Since the RRAM is non-volatile, address accesses can be intermittent and therefore stay undetected from system-level techniques that can identify continuous hammering as a possible security threat. We also present three reset techniques to de-activate the triggers. The resulting static/dynamic power overhead and maximum area overhead incurred by the proposed ENTTs are 104.24 μW/0.426 μW and 9.15 μm2, respectively in PTM 65 nm technology. ENTTs are effective against contemporary Trojan detection techniques and system level protocols. We also propose countermeasures to detect ENTT during the test phase and/or prevent fault-injection attacks during deployment.