Nalini Venkatasubramanian

Praveen Venkateswaran, Vatche Isahagian, Vinod Muthusamy et al.

Existing federated learning models that follow the standard risk minimization paradigm of machine learning often fail to generalize in the presence of spurious correlations in the training data. In many real-world distributed settings, spurious correlations exist due to biases and data sampling issues on distributed devices or clients that can erroneously influence models. Current generalization approaches are designed for centralized training and attempt to identify features that have an invariant causal relationship with the target, thereby reducing the effect of spurious features. However, such invariant risk minimization approaches rely on apriori knowledge of training data distributions which is hard to obtain in many applications. In this work, we present a generalizable federated learning framework called FedGen, which allows clients to identify and distinguish between spurious and invariant features in a collaborative manner without prior knowledge of training distributions. We evaluate our approach on real-world datasets from different domains and show that FedGen results in models that achieve significantly better generalization and can outperform the accuracy of current federated learning approaches by over 24%.

Rahul Atul Bhope, K. R. Jayaram, Nalini Venkatasubramanian et al.

This paper presents the design and implementation of FLIPS, a middleware system to manage data and participant heterogeneity in federated learning (FL) training workloads. In particular, we examine the benefits of label distribution clustering on participant selection in federated learning. FLIPS clusters parties involved in an FL training job based on the label distribution of their data apriori, and during FL training, ensures that each cluster is equitably represented in the participants selected. FLIPS can support the most common FL algorithms, including FedAvg, FedProx, FedDyn, FedOpt and FedYogi. To manage platform heterogeneity and dynamic resource availability, FLIPS incorporates a straggler management mechanism to handle changing capacities in distributed, smart community applications. Privacy of label distributions, clustering and participant selection is ensured through a trusted execution environment (TEE). Our comprehensive empirical evaluation compares FLIPS with random participant selection, as well as three other "smart" selection mechanisms - Oort, TiFL and gradient clustering using two real-world datasets, two benchmark datasets, two different non-IID distributions and three common FL algorithms (FedYogi, FedProx and FedAvg). We demonstrate that FLIPS significantly improves convergence, achieving higher accuracy by 17 - 20 % with 20 - 60 % lower communication costs, and these benefits endure in the presence of straggler participants.

Rahul Atul Bhope, K. R. Jayaram, Vinod Muthusamy et al.

Despite significant costs from retrieving and processing high-fidelity visual inputs, most multimodal vision-language systems operate at fixed fidelity levels. We introduce VOILA, a framework for Value-Of-Information-driven adaptive fidelity selection in Visual Question Answering (VQA) that optimizes what information to retrieve before model execution. Given a query, VOILA uses a two-stage pipeline: a gradient-boosted regressor estimates correctness likelihood at each fidelity from question features alone, then an isotonic calibrator refines these probabilities for reliable decision-making. The system selects the minimum-cost fidelity maximizing expected utility given predicted accuracy and retrieval costs. We evaluate VOILA across three deployment scenarios using five datasets (VQA-v2, GQA, TextVQA, LoCoMo, FloodNet) and six Vision-Language Models (VLMs) with 7B-235B parameters. VOILA consistently achieves 50-60% cost reductions while retaining 90-95% of full-resolution accuracy across diverse query types and model architectures, demonstrating that pre-retrieval fidelity selection is vital to optimize multimodal inference under resource constraints.

Rahul Atul Bhope, Praveen Venkateswaran, K. R. Jayaram et al. · uw

Developers using LLMs and LLM-based agents in their applications have provided plenty of anecdotal evidence that in-context-learning (ICL) is fragile. In this paper, we show that in addition to the quantity and quality of examples, the order in which the in-context examples are listed in the prompt affects the output of the LLM and, consequently, their performance. While prior work has explored improving ICL through dataset-dependent techniques, we introduce OptiSeq, a purely inference-time, dataset-free optimization method that efficiently determines the best example order. OptiSeq leverages log probabilities of LLM-generated outputs to systematically prune the search space of possible orderings and recommend the best order(s) by distinguishing orderings that yield high levels of accuracy and those that underperform. Extensive empirical evaluation on multiple LLMs, datasets, and prompts demonstrate that OptiSeq improves accuracy by 5.5 - 10.5 percentage points across multiple tasks.

Ryan Hildebrant, Rahul Bhope, Sharad Mehrotra et al.

Time series imputation models have traditionally been developed using complete datasets with artificial masking patterns to simulate missing values. However, in real-world infrastructure monitoring, practitioners often encounter datasets where large amounts of data are missing and follow complex, heterogeneous patterns. We introduce DIM-SUM, a preprocessing framework for training robust imputation models that bridges the gap between artificially masked training data and real missing patterns. DIM-SUM combines pattern clustering and adaptive masking strategies with theoretical learning guarantees to handle diverse missing patterns actually observed in the data. Through extensive experiments on over 2 billion readings from California water districts, electricity datasets, and benchmarks, we demonstrate that DIM-SUM outperforms traditional methods by reaching similar accuracy with lower processing time and significantly less training data. When compared against a large pre-trained model, DIM-SUM averages 2x higher accuracy with significantly less inference time.

Rahul Atul Bhope, K. R. Jayaram, Praveen Venkateswaran et al.

Federated Learning (FL) enables collaborative model training across decentralized clients without sharing raw data, yet faces significant challenges in real-world settings where client data distributions evolve dynamically over time. This paper tackles the critical problem of covariate and label shifts in streaming FL environments, where non-stationary data distributions degrade model performance and necessitate a middleware layer that adapts FL to distributional shifts. We introduce ShiftEx, a shift-aware mixture of experts framework that dynamically creates and trains specialized global models in response to detected distribution shifts using Maximum Mean Discrepancy for covariate shifts. The framework employs a latent memory mechanism for expert reuse and implements facility location-based optimization to jointly minimize covariate mismatch, expert creation costs, and label imbalance. Through theoretical analysis and comprehensive experiments on benchmark datasets, we demonstrate 5.5-12.9 percentage point accuracy improvements and 22-95 % faster adaptation compared to state-of-the-art FL baselines across diverse shift scenarios. The proposed approach offers a scalable, privacy-preserving middleware solution for FL systems operating in non-stationary, real-world conditions while minimizing communication and computational overhead.

Nisha Panwar, Shantanu Sharma, Guoxi Wang et al.

Contemporary IoT environments, such as smart buildings, require end-users to trust data-capturing rules published by the systems. There are several reasons why such a trust is misplaced -- IoT systems may violate the rules deliberately or IoT devices may transfer user data to a malicious third-party due to cyberattacks, leading to the loss of individuals' privacy or service integrity. To address such concerns, we propose IoT Notary, a framework to ensure trust in IoT systems and applications. IoT Notary provides secure log sealing on live sensor data to produce a verifiable `proof-of-integrity,' based on which a verifier can attest that captured sensor data adheres to the published data-capturing rules. IoT Notary is an integral part of TIPPERS, a smart space system that has been deployed at the University of California Irvine to provide various real-time location-based services on the campus. We present extensive experiments over realtime WiFi connectivity data to evaluate IoT Notary, and the results show that IoT Notary imposes nominal overheads. The secure logs only take 21% more storage, while users can verify their one day's data in less than two seconds even using a resource-limited device.

Peeyush Gupta, Sharad Mehrotra, Shantanu Sharma et al.

This paper proposes a system, entitled Concealer that allows sharing time-varying spatial data (e.g., as produced by sensors) in encrypted form to an untrusted third-party service provider to provide location-based applications (involving aggregation queries over selected regions over time windows) to users. Concealer exploits carefully selected encryption techniques to use indexes supported by database systems and combines ways to add fake tuples in order to realize an efficient system that protects against leakage based on output-size. Thus, the design of Concealer overcomes two limitations of existing symmetric searchable encryption (SSE) techniques: (i) it avoids the need of specialized data structures that limit usability/practicality of SSE in large scale deployments, and (ii) it avoids information leakages based on the output-size, which may leak data distributions. Experimental results validate the efficiency of the proposed algorithms over a spatial time-series dataset (collected from a smart space) and TPC-H datasets, each of 136 Million rows, the size of which prior approaches have not scaled to.

Peeyush Gupta, Sharad Mehrotra, Nisha Panwar et al.

Contact tracing has emerged as one of the main mitigation strategies to prevent the spread of pandemics such as COVID-19. Recently, several efforts have been initiated to track individuals, their movements, and interactions using technologies, e.g., Bluetooth beacons, cellular data records, and smartphone applications. Such solutions are often intrusive, potentially violating individual privacy rights and are often subject to regulations (e.g., GDPR and CCPR) that mandate the need for opt-in policies to gather and use personal information. In this paper, we introduce Quest, a system that empowers organizations to observe individuals and spaces to implement policies for social distancing and contact tracing using WiFi connectivity data in a passive and privacy-preserving manner. The goal is to ensure the safety of employees and occupants at an organization, while protecting the privacy of all parties. Quest incorporates computationally- and information-theoretically-secure protocols that prevent adversaries from gaining knowledge of an individual's location history (based on WiFi data); it includes support for accurately identifying users who were in the vicinity of a confirmed patient, and then informing them via opt-in mechanisms. Quest supports a range of privacy-enabled applications to ensure adherence to social distancing, monitor the flow of people through spaces, identify potentially impacted regions, and raise exposure alerts. We describe the architecture, design choices, and implementation of the proposed security/privacy techniques in Quest. We, also, validate the practicality of Quest and evaluate it thoroughly via an actual campus-scale deployment at UC Irvine over a very large dataset of over 50M tuples.

Nisha Panwar, Shantanu Sharma, Guoxi Wang et al.

This paper focuses on the new privacy challenges that arise in smart homes. Specifically, the paper focuses on inferring the user's activities -- which may, in turn, lead to the user's privacy -- via inferences through device activities and network traffic analysis. We develop techniques that are based on a cryptographically secure token circulation in a ring network consisting of smart home devices to prevent inferences from device activities, via device workflow, i.e., inferences from a coordinated sequence of devices' actuation. The solution hides the device activity and corresponding channel activities, and thus, preserve the individual's activities. We also extend our solution to deal with a large number of devices and devices that produce large-sized data by implementing parallel rings. Our experiments also evaluate the performance in terms of communication overheads of the proposed approach and the obtained privacy.

Nisha Panwar, Shantanu Sharma, Peeyush Gupta et al.

The growing deployment of Internet of Things (IoT) systems aims to ease the daily life of end-users by providing several value-added services. However, IoT systems may capture and store sensitive, personal data about individuals in the cloud, thereby jeopardizing user-privacy. Emerging legislation, such as California's CalOPPA and GDPR in Europe, support strong privacy laws to protect an individual's data in the cloud. One such law relates to strict enforcement of data retention policies. This paper proposes a framework, entitled IoT Expunge that allows sensor data providers to store the data in cloud platforms that will ensure enforcement of retention policies. Additionally, the cloud provider produces verifiable proofs of its adherence to the retention policies. Experimental results on a real-world smart building testbed show that IoT Expunge imposes minimal overheads to the user to verify the data against data retention policies.

Nisha Panwar, Shantanu Sharma, Guoxi Wang et al.

Contemporary IoT environments, such as smart buildings, require end-users to trust data-capturing rules published by the systems. There are several reasons why such a trust is misplaced --- IoT systems may violate the rules deliberately or IoT devices may transfer user data to a malicious third-party due to cyberattacks, leading to the loss of individuals' privacy or service integrity. To address such concerns, we propose IoT Notary, a framework to ensure trust in IoT systems and applications. IoT Notary provides secure log sealing on live sensor data to produce a verifiable `proof-of-integrity,' based on which a verifier can attest that captured sensor data adheres to the published data-capturing rules. IoT Notary is an integral part of TIPPERS, a smart space system that has been deployed at UCI to provide various real-time location-based services in the campus. IoT Notary imposes nominal overheads for verification, thereby users can verify their data of one day in less than two seconds.

Nisha Panwar, Shantanu Sharma, Sharad Mehrotra et al.

Smart homes are a special use-case of the Internet-of-Things (IoT) paradigm. Security and privacy are two prime concern in smart home networks. A threat-prone smart home can reveal lifestyle and behavior of the occupants, which may be a significant concern. This article shows security requirements and threats to a smart home and focuses on a privacy-preserving security model. We classify smart home services based on the spatial and temporal properties of the underlying device-to-device and owner-to-cloud interaction. We present ways to adapt existing security solutions such as distance-bounding protocols, ISO-KE, SIGMA, TLS, Schnorr, Okamoto Identification Scheme (IS), Pedersen commitment scheme for achieving security and privacy in a cloud-assisted home area network.

Nisha Panwar, Shantanu Sharma, Guoxi Wang et al.

Advances in sensing, networking, and actuation technologies have resulted in the IoT wave that is expected to revolutionize all aspects of modern society. This paper focuses on the new challenges of privacy that arise in IoT in the context of smart homes. Specifically, the paper focuses on preventing the user's privacy via inferences through channel and in-home device activities. We propose a method for securely scheduling the devices while decoupling the device and channels activities. The proposed solution avoids any attacks that may reveal the coordinated schedule of the devices, and hence, also, assures that inferences that may compromise individual's privacy are not leaked due to device and channel level activities. Our experiments also validate the proposed approach, and consequently, an adversary cannot infer device and channel activities by just observing the network traffic.