CENG 793 Advanced Deep Learning (Spring 2018, go here for more recent semesters.)

Instructor:	Gokberk Cinbis
Syllabus:	Please read the syllabus for course details.

Image taken from https://research.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html

Weekly Schedule

Week	Lectures
	Lecture: Introduction, course logistics
	Lecture: Hallmarks of deep learning [slides]
	Lecture: Review: loss functions [slides] Review: optimization [slides] Review: back-propagation [slides] Review: neural-network basics [slides] Review: conv-net basics [slides]
	Lecture: Review: RNNs Paper presentations: New Types of Experiments Reveal that a Neuron Functions as Multiple Independent Threshold Units by Mehmet Ali Akyol. Spiking Neural Networks overview, Fast-Classifying, High-Accuracy Spiking Deep Networks Through Weight and Threshold Balancing and Going Deeper in Spiking Neural Networks: VGG and Residual Architectures all by Yarkin Deniz Cetin.
	Paper presentations: Understanding deep learning requires rethinking generalization by Ahmet Kuzubaşlı. An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling by Ece Caglayan. Nested LSTMs by Ilker Gurcan.
	Paper presentations: Interpretable Explanations of Black Boxes by Meaningful Perturbation by Baris Bulut Gezer.
	Paper presentations: Evolutionary Generative Adversarial Networks by Baris Ozkuslar. [slides] ST-GAN: Spatial Transformer Generative Adversarial Networks for Image Compositing by Burak Kahveci. Neural 3D Mesh Renderer by Selcuk Karakas. Learning Category-Specific Mesh Reconstruction from Image Collections by Kazim Unal.
	Paper presentations: Dynamic Routing Between Capsules and Matrix capsules with EM routing by Acar Erdinc. Exposure: A White-Box Photo Post-Processing Framework by Hilmi Kumdakci. MemGEN: Memory is All You Need by random.
	Progress presentations.
	Paper presentations: BEGAN: Boundary Equilibrium Generative Adversarial Networks by Batuhan Akkaya.
	Paper presentations: Towards High Performance Video Object Detection by Kemal Oksuz. Object Detection with an Aligned Spatial-Temporal Memory by Baris Can Cam. A Read-Write Memory Network for Movie Story Understanding by Baris Caglar Ayyildiz.
	Paper presentations: Group Normalization by Melih Baydar. One Network to Solve Them All --- Solving Linear Inverse Problems using Deep Projection Models by Ogulcan Eryuksel. Brains on Beats by Gunes Sucu.
	Paper presentations: Memory Architectures in Recurrent Neural Network Language Models by Bilgin Coskun. Adversarial PoseNet: A Structure-aware Convolutional Network for Human Pose Estimation by Ozge Unel. Zero-shot Detection by Berkan Demirel.
	No class (will be held on June 1st)
	No class due to academic calendar.
	Final presentations. 9:30-12:30, BMB-4.

Paper presentation guidelines

Each presentation should take min 30 minutes (can take up to 50 minutes).
Please prepare your presentation as a short-lecture that puts the paper into context, rather than presenting a dry summary of the contents of a single paper.
Therefore, please cover important related work (if not already throughly covered in the class) in your presentation, in order to (i) make the presentation accessible for everyone, and, (ii) properly discuss the strengths and weaknesses of the paper compared to related work.
Each presenter is required to ~~send me~~upload a near-complete draft of the slides (as a pdf) 2 days before the presentation date to ODTUclass and send me a simple notification email.
Each presenter is required to re-upload the final version of the slides on the presentation date. Slides will be published on the course webpage.
Please also see week 1 slides for some additional information.
The following will be used for grading: rubric.

Project

Projects can be done individually or as a group (working in groups is encouraged). The goal of the projects is to conduct research at the quality of publications appearing in premier venues.

Project groups Each project group is required to send the names of their members. Due: March 8, 2018, 23:55. [Groups]
Project proposal Each project group is required to submit a one-page project proposal. Due: March 15, 2018, 23:55.
Progress report Each group will prepare a progress report and presentation. Due: April 11, 2018, 23:55.
Final report Each group will prepare a final report and presentation. Report due: ~~May 16~~ May 30, 2018, 23:55.
Please use the ODTUClass submission page. Emailed submissions (unless Moodle is down) and late submissions will not be accepted.
All reports must be prepared using the IEEE double-column conference template. Using LaTeX (or LyX) is recommended. Final report should be around 8 page long.