dasepli / PRML-Spring22-FDU

Course website for PRML Spring 2022 at Fudan University

Home Page:https://dasepli.github.io/PRML-Spring22-FDU/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

PRML-Spring22-FDU

Course website for PRML Spring 2022 at Fudan University (Github Link).

Course Logistics

  • Instructors: Prof. Xipeng Qiu, Prof. Yugang Jiang
  • TAs: Peng Li
  • Time: Monday 6:30 pm - 9:05 pm
  • Venue: H3205 / Online
  • Prerequisites: College Calculus, Linear Algebra, Probability and Statistics, Numerical Optimization and Python Programming
  • TextBooks
    • Neural Network and Deep Learning, Xipeng Qiu, online version
    • Pattern Recognition and Machine Learning, Christopher M. Bishop, online version
  • Grading: 2 assignments with a total 45% weight, and a 50% final project, and 5% for class.
  • Previous Years: Spring 2021 / Spring 2020 / Spring 2019

News

  • [17/05] Final PJ was released, hope you enjoy it!
  • [23/04] Assignment2 was released, hope you enjoy it!
  • [22/04] Number of assignments changed to two.
  • [30/03] Assignment1 was released, hope you enjoy it!
  • [21/02] Welcome to PRML 2022 Spring and assignment0 was released!

Schedule

Date Description Course Materials Events DDLs
21/02 Lec1: Introduction Assignment0: Programming Warmup
28/02 Lec2: Linear Regression Exercise: Linear and Polynomial Regression
07/03 Lec3: K-NN and Decision Tree Exercise: K-NN and Decision Tree
14/03 Lec4: Perceptron and Logistic Regression Exercise: Logistic and Softmax Regression
21/03 Lec5: Kernel Method and SVM
28/03 Lec6: Feedforward Neural Networks Exercise: FFNs
30/03 Assignment1: Machine Learning Meets Fashion out
02/04 Lec7: Convolutional Neural Networks Exercise: CNNs
11/04 Lec8: Recurrent Neural Networks Exercise: RNNs
18/04 Lec9: Attention Mechanism Exercise: Attentions
20/04 Assignment1 due
23/04 Assignment2: Machine Translation and Model Attack out
25/04 Lec10: Unsupervised Learning
02/05 Cancelled for Workers' Day
09/05 Lec11: Model-Independent Machine Learning
16/05 Lec12: Guest Lecture
17/05 Final Project out
22/05 Assignment2 due
23/05 Lec13: Guest Lecture
30/05 Lec14: Guest Lecture
06/06 Lec15: Guest Lecture
13/06 Lec15: TBD
17/05 Final Project due
20/06 Lec15: TBD

Coursework

Guidelines

Different from previous years, we will mainly use the PaddlePaddle AI Studio platform this year for our programming exercises. These exercises will teach you to implement the machine learning models you learned step by step and conduct some basic explorations. In addtion to these exercises, several more interesting assignments and a final project will be designed for you to practice using what you have learned to solve some real-world problems. Notice that we will only grade the assignments and the final project, and leave the exercises just as study materials.

These assignments and the final project will be released both on the e-learning platform and this website. But you have to submit them through the e-learning platform.

Late Policy: Everyone of you have 2 free late days for this semester. You can only use the late days for your assignments but not for the final project report. Once you have exhausted your free late days, we will deduct a late penalty of 25% per additional late day.

Assignment0: Programming Warmup (0%)

This assignment will help you do some basic practices using numpy and paddle.

Assignment1: Machine Learning Meets Fashion (20%)

In this assignment, you will

  • Build your own tiny sklearn-style machine learning library;
  • Hack some practical problems with our processed Fashion-MNIST dataset (CV&Understanding);

DDL of this assignment is 23:59, 20/04, 2022.

Assignment2: Machine Translation and Model Attack (25%)

In this assignment, you will

  • Complete a Chinese to English translation task (NLP&Generation);
  • Attack your builded model;

DDL of this assignment is 23:59, 22/05, 2022.

Final Project: Data Efficient Natural Language Inference (50%)

You are free to choose your favorate project. The default project is a NLI task:

  • Build your NLI model;
  • Training the model with as few as labeled training data;

DDL of the final project is 23:59, 17/06, 2022.

About

Course website for PRML Spring 2022 at Fudan University

https://dasepli.github.io/PRML-Spring22-FDU/


Languages

Language:HTML 80.2%Language:Jupyter Notebook 12.9%Language:TeX 6.8%Language:Python 0.1%