arXiv subjects

Aram Avetisyan

Publications and source records attributed to Aram Avetisyan.

11 recordsLinked to original sources

arXiv2026

Accelerated Methods with Complexity Separation Under Data Similarity for Federated Learning Problems

Heterogeneity within data distribution poses a challenge in many modern federated learning tasks. We formalize it as an optimization problem involving a computationally heavy composite under data similarity. By employing different sets of assumptions, we present several approaches to develop communication-efficient methods. An optimal algorithm is proposed for the convex case. The constructed theory is validated through a series of experiments across various problems.

math.OC↗

arXiv2025

Multi-Track Multimodal Learning on iMiGUE: Micro-Gesture and Emotion Recognition

Micro-gesture recognition and behavior-based emotion prediction are both highly challenging tasks that require modeling subtle, fine-grained human behaviors, primarily leveraging video and skeletal pose data. In this work, we present two multimodal frameworks designed to tackle both problems on the iMiGUE dataset. For micro-gesture classification, we explore the complementary strengths of RGB and 3D pose-based representations to capture nuanced spatio-temporal patterns. To comprehensively represent gestures, video, and skeletal embeddings are extracted using MViTv2-S and 2s-AGCN, respectively. Then, they are integrated through a Cross-Modal Token Fusion module to combine spatial and pose information. For emotion recognition, our framework extends to behavior-based emotion prediction, a binary classification task identifying emotional states based on visual cues. We leverage facial and contextual embeddings extracted using SwinFace and MViTv2-S models and fuse them through an InterFusion module designed to capture emotional expressions and body gestures. Experiments conducted on the iMiGUE dataset, within the scope of the MiGA 2025 Challenge, demonstrate the robust performance and accuracy of our method in the behavior-based emotion prediction task, where our approach secured 2nd place.

cs.CV↗

arXiv2025

Bant: Byzantine Antidote via Trial Function and Trust Scores

Recent advancements in machine learning have improved performance while also increasing computational demands. While federated and distributed setups address these issues, their structures remain vulnerable to malicious influences. In this paper, we address a specific threat: Byzantine attacks, wherein compromised clients inject adversarial updates to derail global convergence. We combine the concept of trust scores with trial function methodology to dynamically filter outliers. Our methods address the critical limitations of previous approaches, allowing operation even when Byzantine nodes are in the majority. Moreover, our algorithms adapt to widely used scaled methods such as Adam and RMSProp, as well as practical scenarios, including local training and partial participation. We validate the robustness of our methods by conducting extensive experiments on both public datasets and private ECG data collected from medical institutions. Furthermore, we provide a broad theoretical analysis of our algorithms and their extensions to the aforementioned practical setups. The convergence guaranties of our methods are comparable to those of classical algorithms developed without Byzantine interference.

cs.LG↗

arXiv2025

Communication-Efficient Federated Learning with Adaptive Number of Participants

Rapid scaling of deep learning models has enabled performance gains across domains, yet it introduced several challenges. Federated Learning (FL) has emerged as a promising framework to address these concerns by enabling decentralized training. Nevertheless, communication efficiency remains a key bottleneck in FL, particularly under heterogeneous and dynamic client participation. Existing methods, such as FedAvg and FedProx, or other approaches, including client selection strategies, attempt to mitigate communication costs. However, the problem of choosing the number of clients in a training round remains extremely underexplored. We introduce Intelligent Selection of Participants (ISP), an adaptive mechanism that dynamically determines the optimal number of clients per round to enhance communication efficiency without compromising model accuracy. We validate the effectiveness of ISP across diverse setups, including vision transformers, real-world ECG classification, and training with gradient compression. Our results show consistent communication savings of up to 30\% without losing the final quality. Applying ISP to different real-world ECG classification setups highlighted the selection of the number of clients as a separate task of federated learning.

cs.LG↗

arXiv2024

Local Methods with Adaptivity via Scaling

The rapid development of machine learning and deep learning has introduced increasingly complex optimization challenges that must be addressed. Indeed, training modern, advanced models has become difficult to implement without leveraging multiple computing nodes in a distributed environment. Distributed optimization is also fundamental to emerging fields such as federated learning. Specifically, there is a need to organize the training process to minimize the time lost due to communication. A widely used and extensively researched technique to mitigate the communication bottleneck involves performing local training before communication. This approach is the focus of our paper. Concurrently, adaptive methods that incorporate scaling, notably led by Adam, have gained significant popularity in recent years. Therefore, this paper aims to merge the local training technique with the adaptive approach to develop efficient distributed learning methods. We consider the classical Local SGD method and enhance it with a scaling feature. A crucial aspect is that the scaling is described generically, allowing us to analyze various approaches, including Adam, RMSProp, and OASIS, in a unified manner. In addition to theoretical analysis, we validate the performance of our methods in practice by training a neural network.

cs.LG↗

arXiv2024

SSSD-ECG-nle: New Label Embeddings with Structured State-Space Models for ECG generation

An electrocardiogram (ECG) is vital for identifying cardiac diseases, offering crucial insights for diagnosing heart conditions and informing potentially life-saving treatments. However, like other types of medical data, ECGs are subject to privacy concerns when distributed and analyzed. Diffusion models have made significant progress in recent years, creating the possibility for synthesizing data comparable to the real one and allowing their widespread adoption without privacy concerns. In this paper, we use diffusion models with structured state spaces for generating digital 10-second 12-lead ECG signals. We propose the SSSD-ECG-nle architecture based on SSSD-ECG with a modified conditioning mechanism and demonstrate its efficiency on downstream tasks. We conduct quantitative and qualitative evaluations, including analyzing convergence speed, the impact of adding positive samples, and assessment with physicians' expert knowledge. Finally, we share the results of physician evaluations and also make synthetic data available to ensure the reproducibility of the experiments described.

eess.SP↗

arXiv2024

Self-Trained Model for ECG Complex Delineation

Electrocardiogram (ECG) delineation plays a crucial role in assisting cardiologists with accurate diagnoses. Prior research studies have explored various methods, including the application of deep learning techniques, to achieve precise delineation. However, existing approaches face limitations primarily related to dataset size and robustness. In this paper, we introduce a dataset for ECG delineation and propose a novel self-trained method aimed at leveraging a vast amount of unlabeled ECG data. Our approach involves the pseudolabeling of unlabeled data using a neural network trained on our dataset. Subsequently, we train the model on the newly labeled samples to enhance the quality of delineation. We conduct experiments demonstrating that our dataset is a valuable resource for training robust models and that our proposed self-trained method improves the prediction quality of ECG delineation.

cs.LG↗

arXiv2023

Deep Neural Networks Generalization and Fine-Tuning for 12-lead ECG Classification

Numerous studies are aimed at diagnosing heart diseases based on 12-lead electrocardiographic (ECG) records using deep learning methods. These studies usually use specific datasets that differ in size and parameters, such as patient metadata, number of doctors annotating ECGs, types of devices for ECG recording, data preprocessing techniques, etc. It is well-known that high-quality deep neural networks trained on one ECG dataset do not necessarily perform well on another dataset or clinical settings. In this paper, we propose a methodology to improve the quality of heart disease prediction regardless of the dataset by training neural networks on a variety of datasets with further fine-tuning for the specific dataset. To show its applicability, we train different neural networks on a large private dataset TIS containing various ECG records from multiple hospitals and on a relatively small public dataset PTB-XL. We demonstrate that training the networks on a large dataset and fine-tuning it on a small dataset from another source outperforms the networks trained only on one small dataset. We also show how the ability of a deep neural networks to generalize allows to improve classification quality of more diseases.

eess.SP↗

arXiv2014

Search for top partners with charge 5e/3

A feasibility study of searches for top partners with charge 5e/3 at the upgraded Large Hadron Collider is performed. The discovery potential and exclusion limits are presented using integrated luminosities of 300 fb$^{-1}$ and 3000 fb$^{-1}$ at center-of-mass energies of 14 and 33 TeV.

hep-ex↗

arXiv2013

Methods and Results for Standard Model Event Generation at $\sqrt{s}$ = 14 TeV, 33 TeV and 100 TeV Proton Colliders (A Snowmass Whitepaper)

This document describes the novel techniques used to simulate the common Snowmass 2013 Energy Frontier Standard Model backgrounds for future hadron colliders. The purpose of many Energy Frontier studies is to explore the reach of high luminosity data sets at a variety of high energy colliders. The generation of high statistics samples which accurately model large integrated luminosities for multiple center-of-mass energies and pile-up environments is not possible using an unweighted event generation strategy -- an approach which relies on event weighting was necessary. Even with these improvements in efficiency, extensive computing resources were required. This document describes the specific approach to event generation using Madgraph5 to produce parton-level processes, followed by parton showering and hadronization with Pythia6, and pile-up and detector simulation with Delphes3. The majority of Standard Model processes for pp interactions at $\sqrt(s)$ = 14, 33, and 100 TeV with 0, 50, and 140 additional pile-up interactions are publicly available.

hep-ex↗

arXiv2013

Snowmass Energy Frontier Simulations

This document describes the simulation framework used in the Snowmass Energy Frontier studies for future Hadron Colliders. An overview of event generation with {\sc Madgraph}5 along with parton shower and hadronization with {\sc Pythia}6 is followed by a detailed description of pile-up and detector simulation with {\sc Delphes}3. Details of event generation are included in a companion paper cited within this paper. The input parametrization is chosen to reflect the best object performance expected from the future ATLAS and CMS experiments; this is referred to as the "Combined Snowmass Detector". We perform simulations of $pp$ interactions at center-of-mass energies $\sqrt{s}=$ 14, 33, and 100 TeV with 0, 50, and 140 additional $pp$ pile-up interactions. The object performance with multi-TeV $pp$ collisions are studied for the first time using large pile-up interactions.

hep-ex↗