{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,12]],"date-time":"2026-01-12T12:28:07Z","timestamp":1768220887610,"version":"3.49.0"},"reference-count":40,"publisher":"MDPI AG","issue":"1","license":[{"start":{"date-parts":[[2026,1,12]],"date-time":"2026-01-12T00:00:00Z","timestamp":1768176000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Serbian Ministry of Science, Technological Development and Innovation","award":["451-03-65\/2024-03\/200123"],"award-info":[{"award-number":["451-03-65\/2024-03\/200123"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Information"],"abstract":"<jats:p>This work investigates the effectiveness of block transform coding (BTC) as a lightweight, training-free quantization strategy for compressing the weights of pretrained deep neural networks. The proposed method applies a rule-based block transform with variance and root mean square error (RMSE)-driven stopping criteria, enabling substantial reductions in bit precision while preserving the statistical structure of convolutional and fully connected layer weights. Unlike uniform 8-bit quantization, BTC dynamically adjusts bit usage across layers and achieves significantly lower distortion for the same compression budget. We evaluate BTC across many pretrained architectures and tabular benchmarks. Experimental results show that BTC consistently reduces storage to 4\u20137.7 bits per weight while maintaining accuracy within 2\u20133% of the 32-bit floating point (FP32) baseline. To further assess scalability and baseline strength, BTC is additionally evaluated on large-scale ImageNet models and compared against a calibrated percentile-based uniform post-training quantization method. The results show that BTC achieves a substantially lower effective bit-width while incurring only a modest accuracy reduction relative to calibration-aware 8-bit quantization, highlighting a favorable compression\u2013accuracy trade-off. BTC also exhibits stable behavior across successive post-training quantization (PTQ) configurations, low quantization noise, and smooth RMSE trends, outperforming na\u00efve uniform quantization under aggressive compression. These findings confirm that BTC provides a scalable, architecture-agnostic, and training-free quantization mechanism suitable for deployment in memory- and computing-constrained environments.<\/jats:p>","DOI":"10.3390\/info17010069","type":"journal-article","created":{"date-parts":[[2026,1,12]],"date-time":"2026-01-12T09:13:01Z","timestamp":1768209181000},"page":"69","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Efficient Quantization of Pretrained Deep Networks via Adaptive Block Transform Coding"],"prefix":"10.3390","volume":"17","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-5358-0032","authenticated-orcid":false,"given":"Milan","family":"Dubljanin","sequence":"first","affiliation":[{"name":"Faculty of Science and Mathematics, University of Pri\u0161tina, Lole Ribara 29, 38220 Kosovska Mitrovica, Serbia"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5868-1764","authenticated-orcid":false,"given":"Stefan","family":"Pani\u0107","sequence":"additional","affiliation":[{"name":"Faculty of Science and Mathematics, University of Pri\u0161tina, Lole Ribara 29, 38220 Kosovska Mitrovica, Serbia"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7692-001X","authenticated-orcid":false,"given":"Milan","family":"Savi\u0107","sequence":"additional","affiliation":[{"name":"Faculty of Science and Mathematics, University of Pri\u0161tina, Lole Ribara 29, 38220 Kosovska Mitrovica, Serbia"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0787-1816","authenticated-orcid":false,"given":"Milan","family":"Dejanovi\u0107","sequence":"additional","affiliation":[{"name":"Faculty of Science and Mathematics, University of Pri\u0161tina, Lole Ribara 29, 38220 Kosovska Mitrovica, Serbia"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0004-2116-9785","authenticated-orcid":false,"given":"Oliver","family":"Popovi\u0107","sequence":"additional","affiliation":[{"name":"Department of Business Studies in Blace, Toplica Academy of Applied Studies, 18400 Prokuplje, Serbia"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2026,1,12]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"1148","DOI":"10.1109\/TCOM.1984.1095973","article-title":"Absolute Moment Block Truncation Coding and Its Application to Color Images","volume":"32","author":"Lema","year":"1987","journal-title":"IEEE Trans. Commun."},{"key":"ref_2","unstructured":"Ronson, J., and DeWitte, J. (1982, January 3\u20135). Adaptive Block Truncation Coding Scheme Using an Edge Following Algorithm. Proceedings of the ICASSP\u201982, IEEE International Conference on Acoustics, Speech, and Signal Processing, Paris, France."},{"key":"ref_3","first-page":"1456","article-title":"A Genetic Algorithm Approach to Multilevel Block Truncation Coding","volume":"82","author":"Chen","year":"1999","journal-title":"IEICE Trans. Fundam. Electron. Commun. Comput. Sci."},{"key":"ref_4","unstructured":"Tsou, C.C., Wu, S.H., and Hu, Y.C. (2005, January 17\u201318). Fast Pixel Grouping Technique for Block Truncation Coding. Proceedings of the WCE 2005 Workshop on Consumer Electronics and Signal Processing, Yunlin, Taiwan."},{"key":"ref_5","unstructured":"Mohamed, D., and Chadi, F. (2011). Image Compression Using Block Transmission Coding. J. Sel. Areas Telecommun. (JSAT), 9\u201313. Available online: https:\/\/scispace.com\/pdf\/image-compression-using-block-truncation-coding-um6vsp9p5k.pdf."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"308","DOI":"10.1093\/comjnl\/37.4.308","article-title":"Compression of Digital Images by Block Truncation Coding: A Survey","volume":"37","author":"Nevalainen","year":"1994","journal-title":"Comput. J."},{"key":"ref_7","first-page":"73","article-title":"Design of Forward Adaptive Uniform Quantizer for Discrete Input Samples for Laplacian Source","volume":"9","year":"2010","journal-title":"Electron. Electr. Eng."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"386","DOI":"10.2478\/v10187-012-0057-z","article-title":"Semi\u2013Logarithmic and Hybrid Quantization of Laplacian Source in Wide Range of Variances","volume":"63","year":"2012","journal-title":"J. Electr. Eng."},{"key":"ref_9","first-page":"334","article-title":"Robust and Switched Nonuniform Scalar Quantization of Gaussian Source in a Wide Dynamic Range of Power","volume":"42","year":"2008","journal-title":"Autom. Remote Control"},{"key":"ref_10","first-page":"967","article-title":"Coding Algorithm Based on Loss Compression Using Scalar Quantization Switching Technique and Logarithmic Companding","volume":"26","year":"2010","journal-title":"J. Inf. Sci. Eng."},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"625","DOI":"10.1007\/s11277-010-9929-3","article-title":"Forward Adaptive Logarithmic Quantizer with New Lossless Coding Method for Laplacian Source","volume":"59","year":"2011","journal-title":"Wirel. Pers. Commun."},{"key":"ref_12","first-page":"135","article-title":"Quantizer Design for Multilevel BTC","volume":"12","year":"2011","journal-title":"Prz. Elektrotechniczny Electr. Rev."},{"key":"ref_13","unstructured":"Esser, S.K., McKinstry, J.L., Bablani, D., Appuswamy, R., and Modha, D.S. (2020, January 26\u201330). Learned Step Size Quantization. Proceedings of the International Conference on Learning Representations (ICLR), Addis Ababa, Ethiopia."},{"key":"ref_14","doi-asserted-by":"crossref","unstructured":"Bhalgat, Y., Lee, J., Nagel, M., Blankevoort, T., and Kwak, N. (2020, January 14\u201319). LSQ+: Improving low-bit quantization through learnable offsets and better initialization. Proceedings of the 2020 IEEE\/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Seattle, WA, USA.","DOI":"10.1109\/CVPRW50498.2020.00356"},{"key":"ref_15","unstructured":"Gong, R., Liu, X., Jiang, S., Li, T., Hu, P., Lin, J., Yu, F., and Yan, J. (November, January 27). Differentiable Soft Quantization: Bridging Full-Precision and Low-Bit Neural Networks. Proceedings of the IEEE\/CVF International Conference on Computer Vision (ICCV), Seoul, Republic of Korea."},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Gholami, A., Kim, S., Dong, Z., Yao, Z., Mahoney, M.W., and Keutzer, K. (2022). A Survey of Quantization Methods for Efficient Neural Network Inference. Low-Power Computer Vision, Chapman and Hall\/CRC.","DOI":"10.1201\/9781003162810-13"},{"key":"ref_17","first-page":"26430","article-title":"BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstruction","volume":"34","author":"Li","year":"2021","journal-title":"Adv. Neural Inf. Process. Syst. (NeurIPS)"},{"key":"ref_18","unstructured":"Wei, Y., Gong, R., Zhang, Q., Li, Y., Li, X., Lin, J., and Liu, Y. (2022, January 18\u201324). QDrop: Randomly Dropping Quantization for Extremely Low-Bit Post-Training Quantization. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, LA, USA."},{"key":"ref_19","unstructured":"Markov, I., Alimohammadi, K., Frantar, E., and Alistarh, D.-A. (2024, January 13\u201316). L-GreCo: Layerwise-Adaptive Gradient Compression for Scalable Deep Learning. Proceedings of the 7th Conference on Machine Learning and Systems (MLSys), Santa Clara, CA, USA."},{"key":"ref_20","unstructured":"Faghri, F., Zhang, R., Markov, I., Alistarh, D., Roy, D., and Ramezani-Kebrya, A. (2020, January 13\u201318). Adaptive Gradient Quantization for Data-Parallel SGD. Proceedings of the 37th International Conference on Machine Learning (ICML), Virtual."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Han, W., Vargaftik, S., Mitzenmacher, M., Karp, B., and Basat, R.B. (2024, January 18\u201319). Beyond Throughput and Compression Ratios: Towards High End-to-End Utility of Gradient Compression. Proceedings of the 23rd ACM Workshop on Hot Topics in Networks (HotNets), Irvine, CA, USA.","DOI":"10.1145\/3696348.3696857"},{"key":"ref_22","unstructured":"Zhang, Z., Li, M., Zhou, Y., Chen, Z., and Xu, C. (2023, January 18\u201321). Evaluating Gradient Quantization for Distributed Deep Learning. Proceedings of the IEEE International Conference on Distributed Computing Systems (ICDCS), Hong Kong, China."},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"758","DOI":"10.3390\/smartcities7020031","article-title":"Video Compression Prototype for Autonomous Vehicles","volume":"7","author":"Wiseman","year":"2024","journal-title":"Smart Cities"},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"144079","DOI":"10.1109\/ACCESS.2025.3597217","article-title":"Image Coding with Data-Driven Fast Transforms Based on Approximate Givens Factorizations","volume":"13","author":"Morawaliyadda","year":"2025","journal-title":"IEEE Access"},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Deni\u0107, B., Peri\u0107, Z., Din\u010di\u0107, M., Peri\u0107, S., Simi\u0107, N., and An\u0111elkovi\u0107, M. (2025). Switched 32-Bit Fixed-Point Format for Laplacian-Distributed Data. Information, 16.","DOI":"10.3390\/info16070574"},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Weede, P., Smietana, P.D., Kuhlenb\u00e4umer, G., Deuschl, G., and Schmidt, G. (2024). Two-Stage Convolutional Neural Network for Classification of Movement Patterns in Tremor Patients. Information, 15.","DOI":"10.3390\/info15040231"},{"key":"ref_27","doi-asserted-by":"crossref","unstructured":"Morawaliyadda, D., and Yahampath, P. (2024, January 19\u201322). Construction of Fast Data-Driven Transforms for Image Compression via Multipath Coordinate Descent on Orthogonal Matrix Manifold. Proceedings of the 2024 Data Compression Conference (DCC), Snowbird, UT, USA.","DOI":"10.1109\/DCC58796.2024.00053"},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"9292","DOI":"10.1109\/TIP.2020.3025203","article-title":"Image coding with data-driven transforms: Methodology, performance and potential","volume":"29","author":"Zhang","year":"2020","journal-title":"IEEE Trans. Image Process."},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"3650","DOI":"10.1109\/TIP.2023.3289064","article-title":"An algorithm for learning orthonormal matrix codebooks for adaptive transform coding","volume":"32","author":"Boragolla","year":"2023","journal-title":"IEEE Trans. Image Process."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"339","DOI":"10.1109\/JSTSP.2020.3034501","article-title":"Nonlinear transform coding","volume":"15","author":"Chou","year":"2021","journal-title":"IEEE J. Sel. Topics Signal Process."},{"key":"ref_31","doi-asserted-by":"crossref","unstructured":"Wu, J., Leng, C., Wang, Y., Hu, Q., and Cheng, J. (2016, January 27\u201330). Quantized Convolutional Neural Networks for Mobile Devices. Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA.","DOI":"10.1109\/CVPR.2016.521"},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"118815","DOI":"10.1109\/ACCESS.2019.2934731","article-title":"Learning a Deep Vector Quantization Network for Image Compression","volume":"7","author":"Lu","year":"2019","journal-title":"IEEE Access"},{"key":"ref_33","unstructured":"Banner, R., Nahshan, Y., Hoffer, E., and Soudry, D. (2019, January 8\u201314). Post Training 4-bit Quantization of Convolutional Networks for Rapid-Deployment. Proceedings of the 33rd International Conference on Neural Information Processing Systems, Vancouver, BC, Canada."},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"1335","DOI":"10.1109\/TCOM.1979.1094560","article-title":"Image Compression Using Block Truncation Coding","volume":"27","author":"Delp","year":"1979","journal-title":"IEEE Trans. Commun."},{"key":"ref_35","unstructured":"Delp, E.J., Saenz, M., and Salama, P. (2000). Block Truncation Coding (BTC). Handbook of Image and Video Processing, Academic Press."},{"key":"ref_36","unstructured":"(2019). IEEE Standard for Floating-Point Arithmetic (Standard No. IEEE Std 754-2019). Available online: https:\/\/standards.ieee.org\/ieee\/754\/6210\/."},{"key":"ref_37","doi-asserted-by":"crossref","unstructured":"Tagliavini, G., Mach, S., Rossi, D., Marongiu, A., and Benini, L. (2018, January 19\u201323). A Transprecision Floating-Point Platform for Ultra-Low Power Computing. Proceedings of the Design, Automation & Test in Europe Conference & Exhibition (DATE), Dresden, Germany.","DOI":"10.23919\/DATE.2018.8342167"},{"key":"ref_38","doi-asserted-by":"crossref","unstructured":"Cattaneo, D., Di Bello, A., Cherubin, S., Terraneo, F., and Agosta, G. (2018, January 29\u201331). Embedded Operating System Optimization through Floating to Fixed Point Compiler Transformation. Proceedings of the 21st Euromicro Conference on Digital System Design (DSD), Prague, Czech Republic.","DOI":"10.1109\/DSD.2018.00042"},{"key":"ref_39","unstructured":"Chenyaofo (2025, December 13). Pytorch-Cifar-Models. Available online: https:\/\/github.com\/chenyaofo\/pytorch-cifar-models."},{"key":"ref_40","doi-asserted-by":"crossref","unstructured":"Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., and Fei-Fei, L. (2009, January 20\u201325). ImageNet: A large-scale hierarchical image database. Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition, Miami, FL, USA.","DOI":"10.1109\/CVPR.2009.5206848"}],"container-title":["Information"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2078-2489\/17\/1\/69\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,12]],"date-time":"2026-01-12T09:29:52Z","timestamp":1768210192000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2078-2489\/17\/1\/69"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,1,12]]},"references-count":40,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,1]]}},"alternative-id":["info17010069"],"URL":"https:\/\/doi.org\/10.3390\/info17010069","relation":{},"ISSN":["2078-2489"],"issn-type":[{"value":"2078-2489","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,1,12]]}}}