{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,22]],"date-time":"2026-04-22T21:04:01Z","timestamp":1776891841351,"version":"3.51.2"},"reference-count":26,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/tdm\/userlicense\/1.0\/"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/legal\/tdmrep-license"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-017"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-012"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-004"}],"content-domain":{"domain":["elsevier.com","sciencedirect.com"],"crossmark-restriction":true},"short-container-title":["Neurocomputing"],"published-print":{"date-parts":[[2026,6]]},"DOI":"10.1016\/j.neucom.2026.133498","type":"journal-article","created":{"date-parts":[[2026,3,27]],"date-time":"2026-03-27T03:52:45Z","timestamp":1774583565000},"page":"133498","update-policy":"https:\/\/doi.org\/10.1016\/elsevier_cm_policy","source":"Crossref","is-referenced-by-count":0,"special_numbering":"C","title":["Fundamental limits of neural network sparsification: Evidence from catastrophic interpretability collapse"],"prefix":"10.1016","volume":"682","author":[{"ORCID":"https:\/\/orcid.org\/0009-0003-1519-8179","authenticated-orcid":false,"given":"Dip","family":"Roy","sequence":"first","affiliation":[]},{"given":"Rajiv","family":"Misra","sequence":"additional","affiliation":[]},{"given":"Sanjay Kumar","family":"Singh","sequence":"additional","affiliation":[]}],"member":"78","reference":[{"key":"10.1016\/j.neucom.2026.133498_bib1","first-page":"1","article-title":"Sparsity in deep learning: pruning and growth for efficient inference and training in neural networks","volume":"22","author":"Hoefler","year":"2021","journal-title":"J. Mach. Learn. Res."},{"key":"10.1016\/j.neucom.2026.133498_bib2","first-page":"129","article-title":"What is the state of neural network pruning?","volume":"2","author":"Blalock","year":"2020"},{"key":"10.1016\/j.neucom.2026.133498_bib3","unstructured":"European Commission, Proposal for a regulation laying down harmonised rules on artificial intelligence (Artificial Intelligence Act), COM\/2021\/206, 2021."},{"key":"10.1016\/j.neucom.2026.133498_bib4","series-title":"Advances in Neural Information Processing Systems","first-page":"1135","article-title":"Learning both weights and connections for efficient neural network","author":"Han","year":"2015"},{"key":"10.1016\/j.neucom.2026.133498_bib5","unstructured":"Cunningham H., et al., Sparse autoencoders find highly interpretable features in language models, arXiv:2309.08600, 2023."},{"key":"10.1016\/j.neucom.2026.133498_bib6","article-title":"Towards monosemanticity: decomposing language models with dictionary learning","author":"Bricken","year":"2023","journal-title":"Transform. Circuits Thread"},{"key":"10.1016\/j.neucom.2026.133498_bib7","series-title":"International Conference on Learning Representations","article-title":"\u03b2-VAE: learning basic visual concepts with a constrained variational framework","author":"Higgins","year":"2017"},{"key":"10.1016\/j.neucom.2026.133498_bib8","series-title":"International Conference on Machine Learning","first-page":"2649","article-title":"Disentangling by factorising","author":"Kim","year":"2018"},{"key":"10.1016\/j.neucom.2026.133498_bib9","series-title":"International Conference on Learning Representations","article-title":"The lottery ticket hypothesis: finding sparse, trainable neural networks","author":"Frankle","year":"2019"},{"key":"10.1016\/j.neucom.2026.133498_bib10","article-title":"Toy models of superposition","author":"Elhage","year":"2022","journal-title":"Transform. Circuits Thread"},{"key":"10.1016\/j.neucom.2026.133498_bib11","series-title":"International Conference on Learning Representations","article-title":"Learning sparse neural networks through L0 regularization","author":"Louizos","year":"2018"},{"key":"10.1016\/j.neucom.2026.133498_bib12","series-title":"International Conference on Machine Learning","first-page":"4646","article-title":"Parameter efficient training of deep convolutional neural networks by dynamic sparse reparameterization","author":"Mostafa","year":"2019"},{"key":"10.1016\/j.neucom.2026.133498_bib13","unstructured":"Sharkey L., et al., Compression and interpretability: analyzing sparse autoencoder dictionaries, arXiv:2403.12901, 2024."},{"key":"10.1016\/j.neucom.2026.133498_bib14","series-title":"Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet","author":"Templeton","year":"2024"},{"key":"10.1016\/j.neucom.2026.133498_bib15","series-title":"International Conference on Learning Representations","article-title":"Scaling and evaluating sparse autoencoders","author":"Gao","year":"2025"},{"key":"10.1016\/j.neucom.2026.133498_bib16","unstructured":"Kulkarni A., Weng T.W., Narayanaswamy V., Liu S., Sakla W.A., Thopalli K., Interpretable and steerable concept bottleneck sparse autoencoders, arXiv:2512.10805, 2025."},{"key":"10.1016\/j.neucom.2026.133498_bib17","unstructured":"Baker Z., Li Y., Analysis of variational sparse autoencoders, arXiv:2509.22994, 2025."},{"key":"10.1016\/j.neucom.2026.133498_bib18","series-title":"Proceedings of the IEEE International Conference on Computer Vision","first-page":"2736","article-title":"Learning efficient convolutional networks through network slimming","author":"Liu","year":"2017"},{"key":"10.1016\/j.neucom.2026.133498_bib19","series-title":"Proceedings of the IEEE International Conference on Computer Vision","first-page":"1389","article-title":"Channel pruning for accelerating very deep neural networks","author":"He","year":"2017"},{"key":"10.1016\/j.neucom.2026.133498_bib20","series-title":"International Conference on Machine Learning","first-page":"2943","article-title":"Rigging the lottery: making all tickets winners","author":"Evci","year":"2020"},{"key":"10.1016\/j.neucom.2026.133498_bib21","article-title":"Isolating sources of disentanglement in variational autoencoders","volume":"31","author":"Chen","year":"2018"},{"key":"10.1016\/j.neucom.2026.133498_bib22","series-title":"International Conference on Machine Learning","first-page":"4114","article-title":"Challenging common assumptions in the unsupervised learning of disentangled representations","author":"Locatello","year":"2019"},{"key":"10.1016\/j.neucom.2026.133498_bib23","article-title":"Stacked capsule graph autoencoders for geometry-aware 3D head pose estimation","volume":"208\u2013209","author":"Mazzia","year":"2021","journal-title":"Comput. Vis. Image Underst."},{"issue":"12","key":"10.1016\/j.neucom.2026.133498_bib24","doi-asserted-by":"crossref","first-page":"5659","DOI":"10.1109\/TIP.2015.2487860","article-title":"Multimodal deep autoencoder for human pose recovery","volume":"24","author":"Hong","year":"2015","journal-title":"IEEE Trans. Image Process."},{"key":"10.1016\/j.neucom.2026.133498_bib25","unstructured":"Matthey L., et al., dSprites: Disentanglement testing sprites dataset. \u3008https:\/\/github.com\/deepmind\/dsprites-dataset\u3009 (2017)."},{"key":"10.1016\/j.neucom.2026.133498_bib26","unstructured":"Burgess C., Kim H., 3D shapes dataset, 2018. \u3008https:\/\/github.com\/deepmind\/3d-shapes\u3009."}],"container-title":["Neurocomputing"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0925231226008957?httpAccept=text\/xml","content-type":"text\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0925231226008957?httpAccept=text\/plain","content-type":"text\/plain","content-version":"vor","intended-application":"text-mining"}],"deposited":{"date-parts":[[2026,4,22]],"date-time":"2026-04-22T20:29:47Z","timestamp":1776889787000},"score":1,"resource":{"primary":{"URL":"https:\/\/linkinghub.elsevier.com\/retrieve\/pii\/S0925231226008957"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6]]},"references-count":26,"alternative-id":["S0925231226008957"],"URL":"https:\/\/doi.org\/10.1016\/j.neucom.2026.133498","relation":{},"ISSN":["0925-2312"],"issn-type":[{"value":"0925-2312","type":"print"}],"subject":[],"published":{"date-parts":[[2026,6]]},"assertion":[{"value":"Elsevier","name":"publisher","label":"This article is maintained by"},{"value":"Fundamental limits of neural network sparsification: Evidence from catastrophic interpretability collapse","name":"articletitle","label":"Article Title"},{"value":"Neurocomputing","name":"journaltitle","label":"Journal Title"},{"value":"https:\/\/doi.org\/10.1016\/j.neucom.2026.133498","name":"articlelink","label":"CrossRef DOI link to publisher maintained version"},{"value":"article","name":"content_type","label":"Content Type"},{"value":"\u00a9 2026 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.","name":"copyright","label":"Copyright"}],"article-number":"133498"}}