arXiv subjects

André Bauer

Publications and source records attributed to André Bauer.

15 recordsLinked to original sources

arXiv2026

Proactive Routing to Interpretable Surrogates with Distribution-Free Safety Guarantees

Model routing determines whether to use an accurate black-box model or a simpler surrogate that approximates it at lower cost or greater interpretability. In deployment settings, practitioners often wish to restrict surrogate use to inputs where its degradation relative to a reference model is controlled. We study proactive (input-based) routing, in which a lightweight gate selects the model before either runs, enabling distribution-free control of the fraction of routed inputs whose degradation exceeds a tolerance τ. The gate is trained to distinguish safe from unsafe inputs, and a routing threshold is chosen via Clopper-Pearson conformal calibration on a held-out set, guaranteeing that the routed-set violation rate is at most α with probability 1-δ. We derive a feasibility condition linking safe routing to the base safe rate π and risk budget α, along with sufficient AUC thresholds ensuring that feasible routing exists. Across 35 OpenML datasets and multiple black-box model families, gate-based conformal routing maintains controlled violation while achieving substantially higher coverage than regression conformal and naive baselines. We further show that probabilistic calibration primarily affects routing efficiency rather than distribution-free validity.

cs.LG↗

arXiv2026

Evaluating Kubernetes Performance for GenAI Inference: From Automatic Speech Recognition to LLM Summarization

As Generative AI (GenAI), particularly inference, rapidly emerges as a dominant workload category, the Kubernetes ecosystem is proactively evolving to natively support its unique demands. This industry paper demonstrates how emerging Kubernetes-native projects can be combined to deliver the benefits of container orchestration, such as scalability and resource efficiency, to complex AI workflows. We implement and evaluate an illustrative, multi-stage use case consisting of automatic speech recognition and summarization. First, we address batch inference by using Kueue to manage jobs that transcribe audio files with Whisper models and Dynamic Accelerator Slicer (DAS) to increase parallel job execution. Second, we address a discrete online inference scenario by feeding the transcripts to a Large Language Model for summarization hosted using llm-d, a novel solution utilizing the recent developments around the Kubernetes Gateway API Inference Extension (GAIE) for optimized routing of inference requests. Our findings illustrate that these complementary components (Kueue, DAS, and GAIE) form a cohesive, high-performance platform, proving Kubernetes' capability to serve as a unified foundation for demanding GenAI workloads: Kueue reduced total makespan by up to 15%; DAS shortened mean job completion time by 36\%; and GAIE working in conjunction with llm-d improved tail Time to First Token latency by up to 90% even under high loads.

cs.ET↗

arXiv2025

Efficient Privacy-Preserving Recommendation on Sparse Data using Fully Homomorphic Encryption

In today's data-driven world, recommendation systems personalize user experiences across industries but rely on sensitive data, raising privacy concerns. Fully homomorphic encryption (FHE) can secure these systems, but a significant challenge in applying FHE to recommendation systems is efficiently handling the inherently large and sparse user-item rating matrices. FHE operations are computationally intensive, and naively processing various sparse matrices in recommendation systems would be prohibitively expensive. Additionally, the communication overhead between parties remains a critical concern in encrypted domains. We propose a novel approach combining Compressed Sparse Row (CSR) representation with FHE-based matrix factorization that efficiently handles matrix sparsity in the encrypted domain while minimizing communication costs. Our experimental results demonstrate high recommendation accuracy with encrypted data while achieving the lowest communication costs, effectively preserving user privacy.

cs.CR↗

arXiv2025

Addressing Reproducibility Challenges in HPC with Continuous Integration

The high-performance computing (HPC) community has adopted incentive structures to motivate reproducible research, with major conferences awarding badges to papers that meet reproducibility requirements. Yet, many papers do not meet such requirements. The uniqueness of HPC infrastructure and software, coupled with strict access requirements, may limit opportunities for reproducibility. In the absence of resource access, we believe that regular documented testing, through continuous integration (CI), coupled with complete provenance information, can be used as a substitute. Here, we argue that better HPC-compliant CI solutions will improve reproducibility of applications. We present a survey of reproducibility initiatives and describe the barriers to reproducibility in HPC. To address existing limitations, we present a GitHub Action, CORRECT, that enables secure execution of tests on remote HPC resources. We evaluate CORRECT's usability across three different types of HPC applications, demonstrating the effectiveness of using CORRECT for automating and documenting reproducibility evaluations.

cs.DC↗

arXiv2025

Decomposing the Time Series Forecasting Pipeline: A Modular Approach for Time Series Representation, Information Extraction, and Projection

With the advent of Transformers, time series forecasting has seen significant advances, yet it remains challenging due to the need for effective sequence representation, memory construction, and accurate target projection. Time series forecasting remains a challenging task, demanding effective sequence representation, meaningful information extraction, and precise future projection. Each dataset and forecasting configuration constitutes a distinct task, each posing unique challenges the model must overcome to produce accurate predictions. To systematically address these task-specific difficulties, this work decomposes the time series forecasting pipeline into three core stages: input sequence representation, information extraction and memory construction, and final target projection. Within each stage, we investigate a range of architectural configurations to assess the effectiveness of various modules, such as convolutional layers for feature extraction and self-attention mechanisms for information extraction, across diverse forecasting tasks, including evaluations on seven benchmark datasets. Our models achieve state-of-the-art forecasting accuracy while greatly enhancing computational efficiency, with reduced training and inference times and a lower parameter count. The source code is available at https://github.com/RobertLeppich/REP-Net.

cs.AI↗

arXiv2025

STEB: In Search of the Best Evaluation Approach for Synthetic Time Series

The growing need for synthetic time series, due to data augmentation or privacy regulations, has led to numerous generative models, frameworks, and evaluation measures alike. Objectively comparing these measures on a large scale remains an open challenge. We propose the Synthetic Time series Evaluation Benchmark (STEB) -- the first benchmark framework that enables comprehensive and interpretable automated comparisons of synthetic time series evaluation measures. Using 10 diverse datasets, randomness injection, and 13 configurable data transformations, STEB computes indicators for measure reliability and score consistency. It tracks running time, test errors, and features sequential and parallel modes of operation. In our experiments, we determine a ranking of 41 measures from literature and confirm that the choice of upstream time series embedding heavily impacts the final score.

cs.LG↗

arXiv2024

Object Proxy Patterns for Accelerating Distributed Applications

Workflow and serverless frameworks have empowered new approaches to distributed application design by abstracting compute resources. However, their typically limited or one-size-fits-all support for advanced data flow patterns leaves optimization to the application programmer -- optimization that becomes more difficult as data become larger. The transparent object proxy, which provides wide-area references that can resolve to data regardless of location, has been demonstrated as an effective low-level building block in such situations. Here we propose three high-level proxy-based programming patterns -- distributed futures, streaming, and ownership -- that make the power of the proxy pattern usable for more complex and dynamic distributed program structures. We motivate these patterns via careful review of application requirements and describe implementations of each pattern. We evaluate our implementations through a suite of benchmarks and by applying them in three substantial scientific applications, in which we demonstrate substantial improvements in runtime, throughput, and memory usage.

cs.DC↗

arXiv2024

Octopus: Experiences with a Hybrid Event-Driven Architecture for Distributed Scientific Computing

Scientific research increasingly relies on distributed computational resources, storage systems, networks, and instruments, ranging from HPC and cloud systems to edge devices. Event-driven architecture (EDA) benefits applications targeting distributed research infrastructures by enabling the organization, communication, processing, reliability, and security of events generated from many sources. To support the development of scientific EDA, we introduce Octopus, a hybrid, cloud-to-edge event fabric designed to link many local event producers and consumers with cloud-hosted brokers. Octopus can be scaled to meet demand, permits the deployment of highly available Triggers for automatic event processing, and enforces fine-grained access control. We identify requirements in self-driving laboratories, scientific data automation, online task scheduling, epidemic modeling, and dynamic workflow management use cases, and present results demonstrating Octopus' ability to meet those requirements. Octopus supports producing and consuming events at a rate of over 4.2 M and 9.6 M events per second, respectively, from distributed clients.

cs.DC↗

arXiv2024

Oil & Water? Diffusion of AI Within and Across Scientific Fields

This study empirically investigates claims of the increasing ubiquity of artificial intelligence (AI) within roughly 80 million research publications across 20 diverse scientific fields, by examining the change in scholarly engagement with AI from 1985 through 2022. We observe exponential growth, with AI-engaged publications increasing approximately thirteenfold (13x) across all fields, suggesting a dramatic shift from niche to mainstream. Moreover, we provide the first empirical examination of the distribution of AI-engaged publications across publication venues within individual fields, with results that reveal a broadening of AI engagement within disciplines. While this broadening engagement suggests a move toward greater disciplinary integration in every field, increased ubiquity is associated with a semantic tension between AI-engaged research and more traditional disciplinary research. Through an analysis of tens of millions of document embeddings, we observe a complex interplay between AI-engaged and non-AI-engaged research within and across fields, suggesting that increasing ubiquity is something of an oil-and-water phenomenon -- AI-engaged work is spreading out over fields, but not mixing well with non-AI-engaged work.

cs.DL↗

arXiv2024

Memory Injections: Correcting Multi-Hop Reasoning Failures during Inference in Transformer-Based Language Models

Answering multi-hop reasoning questions requires retrieving and synthesizing information from diverse sources. Large Language Models (LLMs) struggle to perform such reasoning consistently. Here we propose an approach to pinpoint and rectify multi-hop reasoning failures through targeted memory injections on LLM attention heads. First, we analyze the per-layer activations of GPT-2 models in response to single and multi-hop prompts. We then propose a mechanism that allows users to inject pertinent prompt-specific information, which we refer to as "memories," at critical LLM locations during inference. By thus enabling the LLM to incorporate additional relevant information during inference, we enhance the quality of multi-hop prompt completions. We show empirically that a simple, efficient, and targeted memory injection into a key attention layer can often increase the probability of the desired next token in multi-hop tasks, by up to 424%.

cs.CL↗

arXiv2024

Trillion Parameter AI Serving Infrastructure for Scientific Discovery: A Survey and Vision

Deep learning methods are transforming research, enabling new techniques, and ultimately leading to new discoveries. As the demand for more capable AI models continues to grow, we are now entering an era of Trillion Parameter Models (TPM), or models with more than a trillion parameters -- such as Huawei's PanGu-$Σ$. We describe a vision for the ecosystem of TPM users and providers that caters to the specific needs of the scientific community. We then outline the significant technical challenges and open problems in system design for serving TPMs to enable scientific research and discovery. Specifically, we describe the requirements of a comprehensive software stack and interfaces to support the diverse and flexible requirements of researchers.

cs.LG↗

arXiv2024

Comprehensive Exploration of Synthetic Data Generation: A Survey

Recent years have witnessed a surge in the popularity of Machine Learning (ML), applied across diverse domains. However, progress is impeded by the scarcity of training data due to expensive acquisition and privacy legislation. Synthetic data emerges as a solution, but the abundance of released models and limited overview literature pose challenges for decision-making. This work surveys 417 Synthetic Data Generation (SDG) models over the last decade, providing a comprehensive overview of model types, functionality, and improvements. Common attributes are identified, leading to a classification and trend analysis. The findings reveal increased model performance and complexity, with neural network-based approaches prevailing, except for privacy-preserving data generation. Computer vision dominates, with GANs as primary generative models, while diffusion models, transformers, and RNNs compete. Implications from our performance evaluation highlight the scarcity of common metrics and datasets, making comparisons challenging. Additionally, the neglect of training and computational costs in literature necessitates attention in future research. This work serves as a guide for SDG model selection and identifies crucial areas for future exploration.

cs.LG↗

arXiv2023

Attention Lens: A Tool for Mechanistically Interpreting the Attention Head Information Retrieval Mechanism

Transformer-based Large Language Models (LLMs) are the state-of-the-art for natural language tasks. Recent work has attempted to decode, by reverse engineering the role of linear layers, the internal mechanisms by which LLMs arrive at their final predictions for text completion tasks. Yet little is known about the specific role of attention heads in producing the final token prediction. We propose Attention Lens, a tool that enables researchers to translate the outputs of attention heads into vocabulary tokens via learned attention-head-specific transformations called lenses. Preliminary findings from our trained lenses indicate that attention heads play highly specialized roles in language models. The code for Attention Lens is available at github.com/msakarvadia/AttentionLens.

cs.CL↗

arXiv2023

Telescope: An Automated Hybrid Forecasting Approach on a Level-Playing Field

In many areas of decision-making, forecasting is an essential pillar. Consequently, many different forecasting methods have been proposed. From our experience, recently presented forecasting methods are computationally intensive, poorly automated, tailored to a particular data set, or they lack a predictable time-to-result. To this end, we introduce Telescope, a novel machine learning-based forecasting approach that automatically retrieves relevant information from a given time series and splits it into parts, handling each of them separately. In contrast to deep learning methods, our approach doesn't require parameterization or the need to train and fit a multitude of parameters. It operates with just one time series and provides forecasts within seconds without any additional setup. Our experiments show that Telescope outperforms recent methods by providing accurate and reliable forecasts while making no assumptions about the analyzed time series.

cs.LG↗

arXiv2019

Performance-Feedback Autoscaling with Budget Constraints for Cloud-based Workloads of Workflows

The growing popularity of workflows in the cloud domain promoted the development of sophisticated autoscaling policies that allow automatic allocation and deallocation of resources. However, many state-of-the-art autoscaling policies for workflows are mostly plan-based or designed for batches (ensembles) of workflows. This reduces their flexibility when dealing with workloads of workflows, as the workloads are often subject to unpredictable resource demand fluctuations. Moreover, autoscaling in clouds almost always imposes budget constraints that should be satisfied. The budget-aware autoscalers for workflows usually require task runtime estimates to be provided beforehand, which is not always possible when dealing with workloads due to their dynamic nature. To address these issues, we propose a novel Performance-Feedback Autoscaler (PFA) that is budget-aware and does not require task runtime estimates for its operation. Instead, it uses the performance-feedback loop that monitors the average throughput on each resource type. We implement PFA in the popular Apache Airflow workflow management system, and compare the performance of our autoscaler with other two state-of-the-art autoscalers, and with the optimal solution obtained with the Mixed Integer Programming approach. Our results show that PFA outperforms other considered online autoscalers, as it effectively minimizes the average job slowdown by up to 47% while still satisfying the budget constraints. Moreover, PFA shows by up to 76% lower average runtime than the competitors.

cs.DC↗