DS407
Modern Deep Learning
Faculty
Radoslav Neychev
Harbour.Space AI Track Director, Girafe-ai founder
Course length
Duration
Total hours
Credits
Language
Course type
Fee for single course
Fee for degree students
Skills you’ll learn
Overview
Recent progress in the AI field has been driven by advances in Deep Learning over the last decade, with neural networks powering modern systems across various modalities: text (NLP), vision (CV), audio, graphs, etc.
This course revisits the fundamentals of Deep Learning and focuses on the general approaches and techniques which allow neural networks to work extremely well in different domains, highlighting selection of important achievements in the field, especially featuring NLP and RL due to the wide usage of LLMs.
This course enhances both practical skills and theoretical knowledge and aims to help the students deeply understand modern approaches in DL and be able to work on their own in the AI field.
This course accompanies the Master’s Machine Learning course (Module 5).
Programming assignments will be implemented in Python 3. The PyTorch framework will be used for deep learning practice.
Learning highlights
- Learn to apply deep learning techniques in practice.
- Get familiar with both fundamental and most recent approaches in Deep Learning, focusing on NLP and RL.
- Learn about large language models (LLMs) and how to apply them to different products.
- Get ready to face real-world problems and tackle them with Deep Learning.
- Gain essential experience with the modern AI frameworks.
Course outline
15 classes
Deep Learning Fundamentals recap
Likelihood maximisation, its relation to cross-entropy, backpropagation mechanism, vanishing gradient problem and why regularisation matters.
Informative Embeddings and where to find them
Embeddings from backbones (CNNs, RNNs), word embeddings. Contrastive learning paradigm.
Attention mechanism
(Self-)Attention mechanism foundation. How it allows to capture similarity in different domains: texts, audio, images, graphs, etc.
Tribute to Transformer architecture. Contextual embeddings.
Transformer architecture. Contextual embeddings: ELmO, BERT and beyond.
Almost Large Language Models
Through Machine Translation to Language Modeling. GPT family. Emerging capabilities of GPT-2.
Modern LLMs training
Basic steps of modern large (and small) LMs training. How to train your own LLM from scratch (and maybe not to).
PEFT: LLMs (and other NNs) efficient fine-tuning
How (and why) not to train from scratch. P-tuning, LoRA, etc.
Midterm Test
Q&A, Discussion
Introduction to Reinforcement Learning
Reinforcement learning problem statement. How to approach black-box optimisation.
Model-free and model-based methods in RL
Value function, Q-function. Q-learning.
Policy Gradient Methods
REINFORCE algorithm. Policy gradient main idea. On- and off-policy learning.
Baselines. GRPO
Actor critic, A2C. Baselines outside of simulation. GRPO and its successors.
Alignment. RL in Language Modeling and other domains
RLHF, RLAIF. DPO and other ways to teach LLM to behave.
Agentic Systems
Tools usage in the LLM paradigm. RAG. Orchestration.
Final exam
Extra topics and Q&A session.
Prerequisites
Master’s Machine Learning course or equivalent, e.g., Introduction to Deep Learning and Computer Vision course.
Python programming experience, PyTorch basics.
At least basic knowledge of Linear Algebra, Probability Theory, Optimisation.
Methodology
The course consists of interactive lectures, practical and Q&A sessions, practical home assignments and theoretical tests.
Grading
Radoslav Neychev is a data scientist with focus on Deep Learning and Reinforcement Learning techniques. He has worked on variety of research (CERN LHCb, MIPT Machine Intelligence Lab, CC RAS) and industrial projects (Yandex, RaiffeisenBank) in different domains vary from particle identification problem to fraudulent transactions detection.
Radoslav graduated from Moscow Institute of Physics and Technology, majoring in Applied Mathematics and Machine Learning. Radoslav is reading lectures and organising practical classes at Russian top-tier universities, tech companies and summer schools.
See full profileApply for this course
Modern Deep Learning
by Radoslav Neychev
Total hours
45 Hours
Dates
Feb 02 - Feb 20, 2026
Fee for single course
€1500
Fee for degree students
€750
How to secure your spot
Complete the form below to kickstart your application
Schedule your Harbour.Space interview
If successful, get ready to join us on campus
FAQ
Will I receive a certificate after completion?
Yes. Upon completion of the course, you will receive a certificate signed by the director of the program your course belonged to.
Do I need a visa?
This depends on your case. Please check with the Spanish or Thai consulate in your country of residence about visa requirements. We will do our part to provide you with the necessary documents, such as the Certificate of Enrollment.
Can I get a discount?
Yes. The easiest way to enroll in a course at a discounted price is to register for multiple courses. Registering for multiple courses will reduce the cost per individual course. Please ask the Admissions Office for more information about the other kinds of discounts we offer and what you can do to receive one.