Top 5 Machine Learning Papers in Q3 2017

Here are the most significant machine learning papers from the third quarter of 2017:

1. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Authors: Devlin et al. Key Contribution: Introduced BERT, a pre-trained language model that achieved state-of-the-art results on multiple NLP tasks by using bidirectional context.

2. Deep Learning Scaling is Predictable, Empirically

Authors: Hestness et al. Key Contribution: Provided empirical evidence for power-law scaling in deep learning, helping researchers better understand and predict model performance.

3. Learning to Learn by Gradient Descent by Gradient Descent

Authors: Andrychowicz et al. Key Contribution: Introduced the concept of learning optimizers through meta-learning, showing how neural networks can learn optimization algorithms.

4. Understanding Black-box Predictions via Influence Functions

Authors: Koh & Liang Key Contribution: Developed a method to understand model predictions by identifying influential training examples, improving model interpretability.

5. Neural Architecture Search with Reinforcement Learning

Authors: Zoph & Le Key Contribution: Demonstrated how reinforcement learning could be used to automatically design neural network architectures.

Note: This is a draft post. The content will be expanded with more detailed analysis and implementation details.