Topic outline

  • General

    Collapse all Expand all
    • Forum icon
      Announcements Forum
    • Page icon
      Contact Sheet Page

      Specialization: IABD

      Level: Second Year Master's Degree

      Module: Advanced Deep Learning

      Credit: 6

      Unit: Fundamental 

      Coefficient: 3

       Instructor: Mohamed GOUDJIL

      Email: mohamed.goudjil@univ-dbkm.dz

       

    • Replacement Session for Lab Project Presentations

       

      Students who were unable to attend the initial presentation session of the laboratory projects are notified that a replacement session has been scheduled for Wednesday, 3rd February, in S2 at 11:30 AM.

    • New Assignment Submission

      We have opened a new assignment where you can attach your Activity Lab presentation and code. Please ensure you submit it before January 30th at 00:00.

       

    • Tutorial test (TD Test)

      Please be informed that the Tutorial test (TD Test) will take place on Tuesday, January 28, 2025, at 13:30 PM in Lecture Hall (AP2).

    • حصة تصحيح الاختبار

       

      نعلمكم أنه قد تم برمجت حصة تصحيح الاختبار ليوم الثلاثاء 28 جانفي 2025 على الساعه 14:00 بالمدرج 2. 

       

      أستاذ المادة

    • تأجيل حصة تصحيح الاختبار

       

      نعلمكم أنه قد تم تأجيل حصة تصحيح الاختبار ليوم الاحد 26 جانفي الى يوم لاحق سيتم الاعلان عنه في حينه. 

       

      أستاذ المادة

    • Course Objectifs

      Welcome to our Advanced Deep Learning course! This course is designed to provide a deep understanding of the most cutting-edge concepts and techniques in the field of deep learning. Through a comprehensive exploration of various advanced topics, we aim to equip students with the knowledge and skills required to tackle complex real-world problems using deep learning methodologies. The objectives of this course are as follows:

      This course is designed to be highly interactive, with a blend of theoretical knowledge and hands-on projects to ensure a well-rounded learning experience. By the end of the course, students will be proficient in applying advanced deep learning techniques to real-world scenarios and will be well-prepared for further research or professional endeavors in the field of deep learning.

    • بوابة التعليم الإلكترونية أسلوب بيداغوجي معتمد رسميا من طرف الوزارة.

       

      نلفت عناية طلبتنا الأكارم أن التعليم عن بعد يعتبر أسلوب تعليمي بيداغوجي معتمد -من طرف الجهات الوصية*- ضمن منظومة التعليم العالي.

      وعليه نرجوا منكم -لزاما- المتابعة الدائمة والتجاوب مع كل المنشورات والتكليفات التي نضعها تباعا في صفحة الدرس على البوابة.

      بالتوفيق للجميع.

       

      *حسب قرار الوزارة الوصية رقم 055 مؤرخ في 21 . 01 . 2021.

       

      أستاذ المادة

       

    • برمجة حصص تعويضية

       

      نعلمكم أنه قد تمت برمجة حصتين تغويضيتين يوم الثلاثاء 17ديسمبر 2024 ابتداءا من الساعه الواحدة ظهرا. (حصص محاضرات)

       

      أستاذ المادة

    • تقييم الاعمال الفصلية

      نلفت انتباهكم بأنه ستكون هناك بعض الاعمال التي يتوجب ارسالها عبر المنصة الالكترونية خلال فترة العطلة، وعليه الرجاء من الجميع التاكد من حساباتهم على المنصة.

      بالتوفيق للجميع

    • التذكير بمواضيع العروض

      للتذكير بأن مواضيع العروض يجب ان تحتوي على شطرين، الأول خاص بالبحث النظري والثاني خاص بالكود البرمجي.

  • Chapter 1: CNN

    cnn

    • File icon
      CNN File

      This chapter delves into the foundational aspects of fully connected neural networks, emphasizing their significance in the realm of computer vision. It explores the transition from manual feature extraction, which relies heavily on domain knowledge, to automated learning methods that streamline the process. The content is primarily derived from the MIT Deep Learning course, providing a comprehensive overview of contemporary techniques and their practical applications

    • File icon
      CNN refs File

  • Chapter v2: RNN

    rnn

    • File icon
      Recurrent Neural Networks (RNNs) File

      In this chapter, we delve into Recurrent Neural Networks (RNNs), a type of neural network particularly well-suited for handling sequential data. RNNs are unique in their ability to maintain a 'memory' of previous inputs through loops within their architecture, which is essential for capturing temporal dependencies. This makes them ideal for various applications such as time series analysis, audio processing, and natural language processing. Specific tasks where RNNs excel include binary. classification, sentiment analysis, image captioning, and machine translation

    • File icon
      RNN refs File

  • Chapter 3: Transfer learning

    transfer learning

    • File icon
      Transfer learning File

      In this chapter, the author discusses the fundamentals of transfer learning, starting with an overview of its basic concepts and formal definitions. Various strategies for implementing transfer learning are explored, alongside practical demonstrations to illustrate how these concepts work in real-world applications. The chapter also delves into the motivation behind using transfer learning and its advantages, particularly in scenarios with limited data availability. Additionally, the text provides a concise review of Convolutional Neural Networks (CNNs), which are often used in transfer learning tasks.

  • Deep Reinforcement Learning

    DRL

    • File icon
      Deep_Reinforcement_Learning File

      In this chapter, we delve into the concept of Deep Reinforcement Learning (DRL). DRL merges reinforcement learning (RL) with deep learning, enabling agents to make decisions and learn policies for complex tasks. Unlike traditional RL, which relies on hand-crafted features, DRL uses neural networks to automatically learn representations. This allows agents to handle high-dimensional state spaces, such as those found in video games or robotic control. Through the combination of deep neural networks and reward-based learning, DRL has achieved remarkable successes, including mastering games like Go and complex robotic manipulations.

  • Topic 5

  • LABS

  • Topic 7