[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-125571-en":3,"doc-seo-125571-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},125571,1099513958762,"Logic","https://ap-avatar.wpscdn.com/avatar/1000023916a998db790?x-image-process=image/resize,m_fixed,w_180,h_180&k=1784791008015729253",8,"Research & Report","Tetra-AML - Automatic Machine Learning via Tensor Networks","Tetra-AML is an Automatic Machine Learning toolbox that automates neural architecture search and hyperparameter optimization using a custom black-box tensor train optimization algorithm (TetraOpt). It further reduces model size through quantization and pruning, strengthened by tensor-network-based compression. The work evaluates a unified benchmark for optimizing neural networks in computer vision, outperforming Bayesian optimization on CIFAR-10, and demonstrates ResNet-18 compression with 14.5× less memory and only 3.2% accuracy loss.","Tetra-AML: Automatic Machine Learning via Tensor Networks  \nA. Naumov, Ar. Melnikov, V. Abronin, F. Oxanichenko, K. Izmailov, M. P􀀍itsch, A. Melnikov, and M. Perelshtein *  \nTerra Quantum AG, Kornhausstrasse 25, 9000 St. Gallen, Switzerland  \narXiv :2303 . 16214v1 [ cs .LG] 28 Mar 2023  \nNeural networks have revolutionized many aspects of society but in the era of huge models with billions of parameters, optimizing and deploying them for commercial applications can require significant computational and financial resources. To address these challenges, we introduce the Tetra-AML toolbox, which automates neural architecture search and hyperparameter optimization via a customdeveloped black-box Tensor train Optimization algorithm, TetraOpt. The toolbox also provides model compression through quantization and pruning, augmented by compression using tensor networks. Here, we analyze a unified benchmark for optimizing neural networks in computer vision tasks and show the superior performance of our approach compared to Bayesian optimization on the CIFAR-10 dataset. We also demonstrate the compression of ResNet-18 neural networks, where we use 14 .5 times less memory while losing just 3 .2% of accuracy. The presented framework is generic, not limited by computer vision problems, supports hardware acceleration (such as with GPUs and TPUs) and can be further extended to quantum hardware and to hybrid quantum machine learning models.  \nINTRODUCTION  \nOver the past decade, neural networks have in􀀍uenced practically every aspect of human society [1] . When the cutting-edge neural network AlexNet triumphed in the prestigious ImageNet challenge in the 2010s [2], the startling journey began. Today, AI systems like DALLE- 2 that can produce realistic visuals and art from a description in natural language are still working toward the same overall goal [3] . While neural networks have grown in value for both people and enterprises, putting them into practice in the context of commercial operations is getting more and more di􀀎cult every year.  \nThe process of deploying a model involves several steps, such as model training, architecture and hyperparameter optimization, and testing, which can be computationally intensive and require signi􀀌cant resources. Furthermore, deploying a large-size model can require signi􀀌cant storage and computation resources, which can increase costs [4] . Additionally, if the model needs to be implemented on a small device with limited memory [5], the model may need to be optimized for size, which can add additional complexity and time to the deployment process. At the same time, the model's high accuracy must be maintained.  \nIn addition, the practice of creating models with greater complexity in order to increase accuracy is still developing. There are about 60 million trainable parameters in AlexNet (2012); GPT-2 contains 1.5 billion parameters (2019); DALLE-2 and ChatGPT contain roughly 3.5 billion and 175 billion parameters, respectively,(2022), with the promise to increase the size by at least two orders of magnitude in the coming years. It is clear that the powerful and complicated models are getting more and more expensive in terms of optimization and deployment [6, 7] .  \n* [mpe@terraquantum.swiss](mpe@terraquantum.swiss)  \nTo address these challenges, we develop an Automatic Machine Learning toolbox based on tensor networks, Tetra-AML. A tensor network is a powerful numerical tool that can advance the solution of the high-dimensional problems [8] . Here we develop the framework that allows us to apply tensor networks for automatic machine learning [9], including an automatic search for the best architectures of models { neural architecture search { with optimal parameters { hyperparameters optimization { with the help of our black-box optimization approach, TetraOpt [10] . Besides, it allows compression of the models via a combination of common quantization techniques and pruning approach [11, 12] augmented b","cbCainkBc1H7sAi8","https://ap.wps.com/l/cbCainkBc1H7sAi8","pdf",15476016,1,5,"English","en",105,"# Introduction\n## Neural Architecture Search and Hyperparameters Optimization","[{\"question\":\"What problems does Tetra-AML address in deploying large neural networks?\",\"answer\":\"It targets the computational and financial burden of architecture search, hyperparameter optimization, and deployment, as well as storage and computation costs for large models.\"},{\"question\":\"How does Tetra-AML perform optimization?\",\"answer\":\"Tetra-AML automates neural architecture search and hyperparameter optimization via TetraOpt, a custom black-box optimizer based on tensor trains.\"},{\"question\":\"What kind of compression does Tetra-AML support, and what are the reported results?\",\"answer\":\"It combines quantization and pruning with tensor-network-based compression. For ResNet-18 it reports 14.5× less memory with only a 3.2% drop in accuracy.\"}]","Tetra-AML - Automatic Machine Learning via Tensor Networks | PDF",1785899941,13,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"tetra-aml-automatic-machine-learning-via-tensor-networks","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/tetra-aml-automatic-machine-learning-via-tensor-networks/125571/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-05",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problems does Tetra-AML address in deploying large neural networks?","Question",{"text":75,"@type":76},"It targets the computational and financial burden of architecture search, hyperparameter optimization, and deployment, as well as storage and computation costs for large models.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does Tetra-AML perform optimization?",{"text":80,"@type":76},"Tetra-AML automates neural architecture search and hyperparameter optimization via TetraOpt, a custom black-box optimizer based on tensor trains.",{"name":82,"@type":73,"acceptedAnswer":83},"What kind of compression does Tetra-AML support, and what are the reported results?",{"text":84,"@type":76},"It combines quantization and pruning with tensor-network-based compression. For ResNet-18 it reports 14.5× less memory with only a 3.2% drop in accuracy.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,109,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":106,"show_sort_weight":107,"slug":108},"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":46,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":21,"slug":137},19,"General","general"]