<div dir="ltr"><p align="left" style="margin-top:0.53cm;margin-bottom:0.26cm;line-height:0.55cm;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font style="font-family:Arial,serif;color:rgb(50,48,69);font-size:12pt">[Dear Colleagues, apologies if you receive multiple copies of this message]</font></p><p align="center" style="margin-bottom:0cm;line-height:0.61cm;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font color="#000080"><span lang="zxx"><span style="text-transform:uppercase"><font color="#323045"><font face="Arial, serif"><font style="font-size:16pt"><span style="letter-spacing:0.3pt"><span lang="en-AE">THE FOURTH INTERNATIONAL WORKSHOP ON DEEP AND TRANSFER LEARNING (<span class="gmail-il">DTL2021</span>)</span></span></font></font></font></span></span></font></p><p align="center" style="margin-bottom:0cm;line-height:0.61cm;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font style="font-size:14pt"><span style="text-transform:uppercase"><font color="#414142"><font face="Arial, serif"><span style="letter-spacing:0.3pt">NOVEMBER 15-17, <span class="gmail-il">2021</span>, TARTU, ESTONIA</span></font></font></span></font></p><p align="center" style="margin-bottom:0.26cm;line-height:0.57cm;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font color="#000080"><span lang="zxx"><u><font color="#58595b"><font style="font-size:14pt"><span lang="en-AE"><a href="http://intelligenttech.org/DTL2021/" target="_blank">http://intelligenttech.org/<span class="gmail-il">DTL2021</span>/</a></span></font></font></u></span></font></p><p align="center" style="margin-top:0.53cm;margin-bottom:0.26cm;line-height:0.55cm;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font color="#323045"><font face="Arial, serif"><font style="font-size:14pt">Colocated with</font></font></font></p><p align="center" style="margin-top:0.53cm;margin-bottom:0.26cm;line-height:0.55cm;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font color="#323045"><font face="Arial, serif"><font style="font-size:14pt">The IEEE International Conference on Intelligent Data Science Technologies and Applications (IDSTA2021)</font></font></font></p><p align="justify" style="margin-bottom:0.26cm;line-height:0.57cm;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><b><font color="#58595b"><font style="font-size:12pt">CALL FOR PAPERS</font></font></b></p><p align="justify" style="margin-bottom:0.26cm;line-height:0.57cm;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font style="font-size:12pt"><font color="#58595b">Deep learning approaches have caused tremendous advances in many areas of computer science. Deep learning is a branch of machine learning where the learning process is done using deep and complex architectures such as recurrent convolutional artificial neural networks. Many computer science applications have utilized deep learning such as computer vision, speech recognition, natural language processing, sentiment analysis, social network analysis, and robotics. The success of deep learning enabled the application of learning models such as reinforcement learning in which the learning process is only done by trial-and-error, solely from actions, rewards or punishments. Deep reinforcement learning comes to create systems that can learn how to adapt in the real world. As deep learning utilizes deep and complex architectures, the learning process usually is time and effort consuming and needs huge labeled data sets. This inspired the introduction of transfer and multi-task learning approaches to better exploit the available data during training and adapt previously learned knowledge to emerging domains, tasks, or applications.</font></font></p><p align="justify"><font color="#58595b"><font style="font-size:12pt">Despite the fact that many research activities are ongoing in these areas, many challenges are still unsolved. This workshop will bring together researchers working on deep learning, working on the intersection of deep learning and reinforcement learning, and/or using transfer learning to simplify deep learning, and it will help researchers with expertise in one of these fields to learn about the others. The workshop also aims to bridge the gap between theories and practices by providing the researchers and practitioners the opportunity to share ideas and discuss and criticize current theories and results. We invite the submission of original papers on all topics related to deep learning, deep reinforcement learning, and transfer and multi-task learning, with special interest in but not limited to:</font></font></p><ul><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Deep Learning for Natural Language Processing</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Deep Learning for Recommender Systems</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Deep learning for computer vision</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Deep learning for systems and networks resource management</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Optimization for Deep Learning</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Deep Reinforcement Learning</font></p><ul><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Deep transfer learning for robots</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Determining rewards for machines</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Machine translation</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Energy consumption issues in deep reinforcement learning</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Deep reinforcement learning for game playing</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Stabilize learning dynamics in deep reinforcement learning</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Scaling up prior reinforcement learning solutions</font></p></li></ul></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Deep Transfer and multi-task learning:</font></p><ul><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">New perspectives or theories on transfer and multi-task learning</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Dataset bias and concept drift</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Transfer learning and domain adaptation</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Multi-task learning</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Feature based approaches</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Instance based approaches</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Deep architectures for transfer and multi-task learning</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Transfer across different architectures, e.g. CNN to RNN</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Transfer across different modalities, e.g. image to text</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Transfer across different tasks, e.g. object recognition and detection</font></p></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Transfer from weakly labeled or noisy data, e.g. Web data</font></p></li></ul></li><li style="margin-left:15px"><p style="margin-bottom:0cm"><font color="#58595b">Datasets, benchmarks, and open-source packages</font></p></li><li style="margin-left:15px"><p><font color="#58595b">Recourse efficient deep learning</font></p></li></ul><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><b><font style="font-size:12pt">FULL PAPER IMPORTANT DATES</font></b></font></p><ul><li style="margin-left:15px"><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><font face="Calibri, serif"><font style="font-size:12pt"><font color="#ff0000">Full paper submission: September 25th, <span class="gmail-il">2021</span></font></font></font></font></p></li><li style="margin-left:15px"><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><font style="font-size:12pt">Full paper acceptance notification: October 5</font><font face="Calibri, serif"><font style="font-size:12pt"><span lang="en-AE">th</span></font></font><font style="font-size:12pt">, <span class="gmail-il">2021</span></font></font></p></li><li style="margin-left:15px"><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><font style="font-size:12pt">Full paper camera-ready submission: October </font><font style="font-size:12pt">20</font><font style="font-size:12pt">th, <span class="gmail-il">2021</span></font></font></p></li></ul><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><b><font style="font-size:12pt">Submission Site:</font></b></font></p><p><font color="#58595b"><font style="font-size:12pt"><a href="https://easychair.org/conferences/?conf=dtl2021" target="_blank">https://easychair.org/conferences/?conf=<span class="gmail-il">dtl2021</span></a></font></font></p><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><b><font style="font-size:12pt">Paper format</font></b></font></p><p><font color="#58595b"><font style="font-size:12pt">Submitted papers (.pdf format) must use <a href="https://www.ieee.org/conferences/publishing/templates.html" target="_blank">the A4 IEEE Manuscript Templates for Conference Proceedings</a>. Please remember to add Keywords to your submission.</font></font></p><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><b><font style="font-size:12pt">Length</font></b></font></p><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><font style="font-size:12pt">Submitted papers may be 6 to 8 pages. Up to two additional pages may be added for references. The reference pages must only contain references. Overlength papers will be rejected without review.</font></font></p><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><b><font style="font-size:12pt">Originality</font></b></font></p><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><font style="font-size:12pt">Papers submitted to DTL must be the original work of the authors. They may not be simultaneously under review elsewhere. Publications that have been peer-reviewed and have appeared at other conferences or workshops may not be submitted to DTL. Authors should be aware that IEEE has a strict policy with regard to plagiarism <a href="https://www.ieee.org/publications/rights/plagiarism/plagiarism-faq.html" target="_blank">https://www.ieee.org/publications/rights/plagiarism/plagiarism-faq.html</a> The authors' prior work must be cited appropriately.</font></font></p><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><b><font style="font-size:12pt">Publication:</font></b></font></p><p style="margin-bottom:0.28cm;line-height:14.04px"><font color="#58595b"><font style="font-size:12pt">All papers that are accepted, registered, and presented in IDSTA2021 and the workshops co-located with it will be submitted to IEEEXplore for possible publication.</font></font></p><p>Best regards,</p><p style="margin-bottom:0cm">IDSTA organising committee</p></div>