武小荷 張宏志 任冬偉 左旺孟 王春宇 劉揚 劉國軍
中心現有教師8人，其中博士生導師5人、教授1人、副教授6人、助理教授1人。研究方向主要包括機器學(xué)習、知識計算及其在計算機視覺(jué)和多模態(tài)任務(wù)理解中的應用。研究中心旨在立足學(xué)科前沿和國家重大需求，匯聚優(yōu)勢力量加強機器學(xué)習理論、模型和應用研究。中心累計完成國家科技部、國家自然科學(xué)基金、省部委基金和企業(yè)橫向項目50余項，獲得省部級科技進(jìn)步獎5項。在CVPR/ICCV/NeurIPS/ECCV等頂級會(huì )議和T-PAMI/IJCV和IEEE Trans.等頂級期刊上發(fā)表論文150余篇，部分成果在國際上產(chǎn)生了較大影響，谷歌學(xué)術(shù)引用24,000余次。
標志性成果：模型引導的深度視覺(jué)計算
針對當前深度學(xué)習在底層視覺(jué)復原與增強和視覺(jué)語(yǔ)義理解中存在的共性和關(guān)鍵瓶頸問(wèn)題，提出融合傳統方法和已有知識/模型發(fā)展新型深度網(wǎng)絡(luò )模型算法的創(chuàng )新研究思路，構建基于模型引導的深度視覺(jué)計算理論與方法體系，在圖像去噪、視覺(jué)復原與增強、視覺(jué)語(yǔ)義理解方面取得了突破性進(jìn)展，對相關(guān)領(lǐng)域的發(fā)展起到了重要的推動(dòng)作用。提出的DnCNN取得顯著(zhù)優(yōu)于傳統方法的去噪性能與計算效率，已成為圖像去噪領(lǐng)域的基準算法。構建了底層視覺(jué)先驗建模、深度去噪網(wǎng)絡(luò )和優(yōu)化算法的有機結合機制，提出了圖像復原網(wǎng)絡(luò )和迭代自適應先驗學(xué)習模型。2018年陜西省科學(xué)技術(shù)一等獎，2020年獲黑龍江省自然科學(xué)一等獎。
Center on Machine Learning Research (CMLR) currently has 8 teachers, including 5 doctoral supervisors, 1 professor, 6 associate professors, and 1 assistant professor. The main research interests include machine learning, knowledge-based computing and their applications in computer vision and multi-modal understanding task. Towards the frontiers of ML disciplines and national requirements, CMLR aims at strengthening the studies on machine learning theories, models, and applications. The center has completed more than 50 research projects, including the funding of Ministry of Science and Technology, the National Natural Science Foundation of China (NSFC), provincial and ministerial funds and enterprise projects. CMLR also gained 5 provincial and ministerial scientific and technological progress awards. More than 150 papers have been published in top conferences such as CVPR/ICCV/NeurIPS/ECCV and top journals such as T-PAMI/IJCV and IEEE Transactions, and received more than 24,000 Google scholar citations.
