tuanlda78202/mlr

Machine Learning Research

Machine Learning Research

1. Project Topic

1.1. Existing research

• Connected Papers

• Papers with Code

• Comprehensive Study

• Google Scholar

• arXiv

• Sci-Hub

• LibGen

• Two Minute Papers

• SOTA Report 2021

• Science Direct

• Towards Data Science

• ACL NLP

• CVF

• ACM

1.2. Datasets and Tasks

• Huggingface Datasets

• Kaggle Datasets

• Wikipedia has a list of machine learning text datasets, tabulated with useful information such as dataset size

• Datahub has lots of datasets, though not all of it is Machine Learning focused

• Microsoft Research has a collection of datasets (look under the ‘Dataset directory’ tab)

• SOTA NLP:

  • nlpprogress

  • gluebenchmark

  • conll

• A small list of well-known standard datasets for common NLP tasks:

  • machinelearningmastery

• An alphabetical list of free or public domain text datasets:

  • nlpdatasets

• Datasets for machine translation:

  • statmt

• Syntactic corpora for many languages:

  • universaldependencies

• A script to search arXiv papers for a keyword, and extract important information such as performance metrics on a task:

  • huyenchip

1.13. Tools

• Overleaf for writing AI research papers with LaTeX.
• Mendeley for reference BibTeX.

2. Project Advice

Processing Data

• StanfordNLP, a Python library providing tokenization, tagging, parsing, and other capabilities.

• Software from the Stanford NLP Group

• NLTK, a lightweight Natural Language Toolkit package in Python.

• spaCy, another Python package that can do preprocessing, but also includes neural models (e.g. Language Models)

3. Top Tiers ML&AI Conferences

• Machine Learning

  • NeurIPS
    • Neural Information Processing Systems (formerly abbreviated NIPS). NeurIPS has gotten huge over the past few years as AI has become so important. Has a focus on neural networks, but not exclusively.
  • ICML
    • International Conference on Machine Learning. Has a general machine learning focus.
  • ICLR
    • International Conference on Learning Representations. ICLR was really the first conference focused on deep learning. It’s called “learning representations” because the motivation behind deep learning is to automatically learn higher-level features, or representations, that summarize data in useful ways. Deep Learning describes the structure of our current best solution to the problem of learning these representations.
  • AISTATS

• Computer Vision

  • CVPR
    • Computer Vision and Pattern Recognition.
  • ICCV
    • International Conference on Computer Vision.
  • ECCV

• Natural Language Processing

  • ACL
  • NAACL
  • EMNLP

• Data

  • KDD
  • CIKM
  • ICDM
  • SDM
  • PAKDD
  • PKDD/ECML
  • RECSYS
  • SIGIR
  • WWW
  • WSDM

• Artificial Intelligence

  • AAAI
    • Association for the Advancement of Artificial Intelligence. AAAI is a little more applications focused, and a little less theoretical than some of the other AI conferences.
  • ICANN
  • IJCAI
  • UAI

4. Courses

• Maths for Machine Learning
• Applied Data Science with Python Specialization
• TensorFlow Developer Professional Certificate
• Google Data Analytics Professional Certificate
• CS224W: Machine Learning with Graphs
• CS230: Deep Learning Specialization
• CS231n: Convolutional Neural Networks for Visual Recognition