-
From N-grams to CodeX (Part 2-NMT, Attention, Transformer)
Many language understanding tasks that were almost infeasible to solve only few years ago emerge as ready-to-use products nowadays. This series serves as an overview to language models, the cornerstone for almost all language tasks.
-
From N-grams to CodeX (Part 1 - N-grams -> RNN)
Many language understanding tasks that were almost infeasible to solve only few years ago emerge as ready-to-use products nowadays. This series serves as an overview to language models, the cornerstone for almost all language tasks.
-
Neural Implicit Representations for 3D Shapes and Scenes
In recent years there is an explosion of neural implicit representations that helps solve computer graphic tasks. In this post, I focus on their applicability to three different tasks - shape representation, novel view synthesis, and image-based 3D reconstruction.
-
Deep Learning Optimization Theory - Trajectory Analysis of Gradient Descent
A prominent approach in the study of deep learning theory in recent years has been analyzing the trajectories followed by gradient descent. This post is an introduction to this approach & my path to understanding it a little better.
-
Deep Learning Optimization Theory - Introduction
Understanding the thoery of optimization in deep learning is crucial to enable progress. This post provides an introduction to it.
-
From A* to MARL (Part 5- Multi-Agent Reinforcement Learning)
An intuitive high-level overview of the connection between AI planning theory to current Reinforcement Learning research for multi-agent systems. This part (finally!) focus on reinforcement learning (RL) and multi-agent RL.
-
From A* to MARL (Part 4 - Planning Under Uncertainty & Partial Observability)
An intuitive high-level overview of the connection between AI planning theory to current Reinforcement Learning research for multi-agent systems. This part focus on POMDPs and planning under partial observability.
-
From A* to MARL (Part 3 - Planning Under Uncertainty)
An intuitive high-level overview of the connection between AI planning theory to current Reinforcement Learning research for multi-agent systems. This part focus on MDP and planning under uncertainty.
-
From A* to MARL (Part 2 - AI Planning)
An intuitive high-level overview of the connection between AI planning theory to current Reinforcement Learning research for multi-agent systems. This part focus on AI Planning.
-
From A* to MARL (Part 1 - MAPF)
An intuitive high-level overview of the connection between AI planning theory to current Reinforcement Learning research for multi-agent systems. This part focuses on the path finding problem and its multi-agent generalization.