Multi-Task Learning with Deep Neural Networks

A baby learns general motor skills while learning to walk, augments, and uses later in life to perform more complex tasks such as playing soccer

What is Multi-Task Learning?

What does the paper focus on?

Expanded Multi-Task Learning Methods

What are different Multi-Task Learning Architectures?

Multi-Task Architectures

Task Domain Architectures

Architecture for TCDCN (Zhang et al., 2014)
The network architecture of (Liu et al., 2015)

Multi-Modal Architectures

OmniNet architecture proposed in (Pramanik et al., 2019)

Learned Architectures

Branched sharing architecture proposed in (Lu et al., 2017)
AdaShare is a modular sharing scheme proposed by (Sun et al., 2019)

Conditional Architectures

Example Neural Module Network execution (Andreas et al., 2016)

What are different Multi-Task Optimization methods?

Multi-Task Optimization Methods

Loss Weighting

Kendal et. al., 2017

Task Scheduling

Task scheduling visualization from (Sharma et al., 2017)

Gradient Modulation

Multi-task GREAT model (Sinha et al., 2018)

Knowledge Distillation

Two architectures from the Distral framework for RL (Teh et al., 2017)

What is Multi-Task Relationship Learning?

Multi-Task Relationship Learning

Grouping Tasks

An example partitioning of a group of tasks into clusters with positive transfer (Standley et al., 2019)

Transfer Relationship Learning

Task taxonomies for a collection of computer vision tasks as computed in Taskonomy (Zamir et al., 2018)

What are different Multi-Task Benchmarks?

Computer Vision Benchmarks

Natural Language Processing Benchmarks

Reinforcement Learning Benchmarks

Multi-Modal Benchmarks

Conclusion

--

--

--

Seasoned R&D EDA, Data Science Enthusiast, Cultural Explorer

Love podcasts or audiobooks? Learn on the go with our new app.

Recommended from Medium

Fast — CNN: Substitution of Convolution layers with FFT layers.

Learn IBM Watson Discovery During the COVID-19 Pandemic

Tuning parameters of SVM: Kernel, Regularization, Gamma and Margin.

Announcing the Kite Integration for Spyder

Applications of Deep Learning in Bioinformatics

Better Data Loading: 20x PyTorch Speed-Up for Tabular Data

Weekly Digest for Data Science and AI: Python and R (Volume 11)

AnimeGAN2-PyTorch Make a photo into an anime

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Abhishek Bais

Abhishek Bais

Seasoned R&D EDA, Data Science Enthusiast, Cultural Explorer

More from Medium

Confusion Matrix without Confused

Finding the instance of a problem using Naive Bayesian Classifier.

Evaluation Metrics for Classification- beyond Accuracy

Predicting Patient No-Shows