arXiv subjects

Ramin Yahyapour

Publications and source records attributed to Ramin Yahyapour.

8 recordsLinked to original sources

arXiv2025

HyDRA: Hierarchical and Dynamic Rank Adaptation for Mobile Vision Language Model

Vision Language Models (VLMs) have undergone significant advancements, particularly with the emergence of mobile-oriented VLMs, which offer a wide range of application scenarios. However, the substantial computational requirements for training these models present a significant obstacle to their practical application. To address this issue, Low-Rank Adaptation (LoRA) has been proposed. Nevertheless, the standard LoRA with a fixed rank lacks sufficient capability for training mobile VLMs that process both text and image modalities. In this work, we introduce HyDRA, a parameter-efficient fine-tuning framework designed to implement hierarchical and dynamic rank scheduling for mobile VLMs. This framework incorporates two essential optimization strategies: (1) hierarchical optimization, which involves a coarse-grained approach that assigns different ranks to various layers, as well as a fine-grained method that adjusts ranks within individual layers, and (2) dynamic adjustment, which employs an end-to-end automatic optimization using a lightweight performance model to determine and adjust ranks during the fine-tuning process. Comprehensive experiments conducted on popular benchmarks demonstrate that HyDRA consistently outperforms the baseline, achieving a 4.7\% improvement across various model sizes without increasing the number of trainable parameters. In some tasks, it even surpasses full-parameter fine-tuning.

cs.LG↗

arXiv2025

Benchmarking pig detection and tracking under diverse and challenging conditions

To ensure animal welfare and effective management in pig farming, monitoring individual behavior is a crucial prerequisite. While monitoring tasks have traditionally been carried out manually, advances in machine learning have made it possible to collect individualized information in an increasingly automated way. Central to these methods is the localization of animals across space (object detection) and time (multi-object tracking). Despite extensive research of these two tasks in pig farming, a systematic benchmarking study has not yet been conducted. In this work, we address this gap by curating two datasets: PigDetect for object detection and PigTrack for multi-object tracking. The datasets are based on diverse image and video material from realistic barn conditions, and include challenging scenarios such as occlusions or bad visibility. For object detection, we show that challenging training images improve detection performance beyond what is achievable with randomly sampled images alone. Comparing different approaches, we found that state-of-the-art models offer substantial improvements in detection quality over real-time alternatives. For multi-object tracking, we observed that SORT-based methods achieve superior detection performance compared to end-to-end trainable models. However, end-to-end models show better association performance, suggesting they could become strong alternatives in the future. We also investigate characteristic failure cases of end-to-end models, providing guidance for future improvements. The detection and tracking models trained on our datasets perform well in unseen pens, suggesting good generalization capabilities. This highlights the importance of high-quality training data. The datasets and research code are made publicly available to facilitate reproducibility, re-use and further development.

cs.CV↗

arXiv2025

Re-Representation in Sentential Relation Extraction with Sequence Routing Algorithm

Sentential relation extraction (RE) is an important task in natural language processing (NLP). In this paper we propose to do sentential RE with dynamic routing in capsules. We first show that the proposed approach outperform state of the art on common sentential relation extraction datasets Tacred, Tacredrev, Retacred, and Conll04. We then investigate potential reasons for its good performance on the mentioned datasets, and yet low performance on another similar, yet larger sentential RE dataset, Wikidata. As such, we identify noise in Wikidata labels as one of the reasons that can hinder performance. Additionally, we show associativity of better performance with better re-representation, a term from neuroscience referred to change of representation in human brain to improve the match at comparison time. As example, in the given analogous terms King:Queen::Man:Woman, at comparison time, and as a result of re-representation, the similarity between related head terms (King,Man), and tail terms (Queen,Woman) increases. As such, our observation show that our proposed model can do re-representation better than the vanilla model compared with. To that end, beside noise in the labels of the distantly supervised RE datasets, we propose re-representation as a challenge in sentential RE.

cs.CL↗

arXiv2025

Workflow-Driven Modeling for the Compute Continuum: An Optimization Approach to Automated System and Workload Scheduling

The convergence of IoT, Edge, Cloud, and HPC technologies creates a compute continuum that merges cloud scalability and flexibility with HPC's computational power and specialized optimizations. However, integrating cloud and HPC resources often introduces latency and communication overhead, which can hinder the performance of tightly coupled parallel applications. Additionally, achieving seamless interoperability between cloud and on-premises HPC systems requires advanced scheduling, resource management, and data transfer protocols. Consequently, users must manually allocate complex workloads across heterogeneous resources, leading to suboptimal task placement and reduced efficiency due to the absence of an automated scheduling mechanism. To overcome these challenges, we introduce a comprehensive framework based on rigorous system and workload modeling for the compute continuum. Our method employs established tools and techniques to optimize workload mapping and scheduling, enabling the automatic orchestration of tasks across both cloud and HPC infrastructures. Experimental evaluations reveal that our approach could optimally improve scheduling efficiency, reducing execution times, and enhancing resource utilization. Specifically, our MILP-based solution achieves optimal scheduling and makespan for small-scale workflows, while heuristic methods offer up to 99% faster estimations for large-scale workflows, albeit with a 5-10% deviation from optimal results. Our primary contribution is a robust system and workload modeling framework that addresses critical gaps in existing tools, paving the way for fully automated orchestration in HPC-compute continuum environments.

cs.DC↗

arXiv2024

Bridging Textual and Tabular Worlds for Fact Verification: A Lightweight, Attention-Based Model

FEVEROUS is a benchmark and research initiative focused on fact extraction and verification tasks involving unstructured text and structured tabular data. In FEVEROUS, existing works often rely on extensive preprocessing and utilize rule-based transformations of data, leading to potential context loss or misleading encodings. This paper introduces a simple yet powerful model that nullifies the need for modality conversion, thereby preserving the original evidence's context. By leveraging pre-trained models on diverse text and tabular datasets and by incorporating a lightweight attention-based mechanism, our approach efficiently exploits latent connections between different data types, thereby yielding comprehensive and reliable verdict predictions. The model's modular structure adeptly manages multi-modal information, ensuring the integrity and authenticity of the original evidence are uncompromised. Comparative analyses reveal that our approach exhibits competitive performance, aligning itself closely with top-tier models on the FEVEROUS benchmark.

cs.CL↗

arXiv2018

Hybrid Job-driven Scheduling for Virtual MapReduce Clusters

It is cost-efficient for a tenant with a limited budget to establish a virtual MapReduce cluster by renting multiple virtual private servers (VPSs) from a VPS provider. To provide an appropriate scheduling scheme for this type of computing environment, we propose in this paper a hybrid job-driven scheduling scheme (JoSS for short) from a tenant's perspective. JoSS provides not only job level scheduling, but also map-task level scheduling and reduce-task level scheduling. JoSS classifies MapReduce jobs based on job scale and job type and designs an appropriate scheduling policy to schedule each class of jobs. The goal is to improve data locality for both map tasks and reduce tasks, avoid job starvation, and improve job execution performance. Two variations of JoSS are further introduced to separately achieve a better map-data locality and a faster task assignment. We conduct extensive experiments to evaluate and compare the two variations with current scheduling algorithms supported by Hadoop. The results show that the two variations outperform the other tested algorithms in terms of map-data locality, reduce-data locality, and network overhead without incurring significant overhead. In addition, the two variations are separately suitable for different MapReduce-workload scenarios and provide the best job performance among all tested algorithms.

cs.DC↗

arXiv2017

A JSON Token-Based Authentication and Access Management Schema for Cloud SaaS Applications

Cloud computing is significantly reshaping the computing industry built around core concepts such as virtualization, processing power, connectivity and elasticity to store and share IT resources via a broad network. It has emerged as the key technology that unleashes the potency of Big Data, Internet of Things, Mobile and Web Applications, and other related technologies, but it also comes with its challenges - such as governance, security, and privacy. This paper is focused on the security and privacy challenges of cloud computing with specific reference to user authentication and access management for cloud SaaS applications. The suggested model uses a framework that harnesses the stateless and secure nature of JWT for client authentication and session management. Furthermore, authorized access to protected cloud SaaS resources have been efficiently managed. Accordingly, a Policy Match Gate (PMG) component and a Policy Activity Monitor (PAM) component have been introduced. In addition, other subcomponents such as a Policy Validation Unit (PVU) and a Policy Proxy DB (PPDB) have also been established for optimized service delivery. A theoretical analysis of the proposed model portrays a system that is secure, lightweight and highly scalable for improved cloud resource security and management.

cs.CR↗

arXiv2017

Reliable Virtual Machine Placement and Routing in Clouds

In current cloud computing systems, when leveraging virtualization technology, the customer's requested data computing or storing service is accommodated by a set of communicated virtual machines (VM) in a scalable and elastic manner. These VMs are placed in one or more server nodes according to the node capacities or failure probabilities. The VM placement availability refers to the probability that at least one set of all customer's requested VMs operates during the requested lifetime. In this paper, we first study the problem of placing at most H groups of k requested VMs on a minimum number of nodes, such that the VM placement availability is no less than $δ$, and that the specified communication delay and connection availability for each VM pair under the same placement group are not violated. We consider this problem with and without Shared-Risk Node Group (SRNG) failures, and prove this problem is NP-hard in both cases. We subsequently propose an exact Integer Nonlinear Program (INLP) and an efficient heuristic to solve this problem. We conduct simulations to compare the proposed algorithms with two existing heuristics in terms of performance. Finally, we study the related reliable routing problem of establishing a connection over at most w link-disjoint paths from a source to a destination, such that the connection availability requirement is satisfied and each path delay is no more than a given value. We devise an exact algorithm and two heuristics to solve this NP-hard problem, and evaluate them via simulations.

cs.DS↗