Added HoMe paper

README.md

@@ -7,6 +7,7 @@ I am trying a new initiative - a-paper-a-week. This repository will hold all tho
 
 * [Poincaré Embeddings for Learning Hierarchical Representations](https://shagunsodhani.in/papers-I-read/Poincare-Embeddings-for-Learning-Hierarchical-Representations)
 * [When Recurrent Models Don’t Need To Be Recurrent](https://shagunsodhani.in/papers-I-read/When-Recurrent-Models-Don-t-Need-To-Be-Recurrent)
+* [HoME - a Household Multimodal Environment](https://shagunsodhani.in/papers-I-read/HoME-a-Household-Multimodal-Environment)
 * [Emergence of Grounded Compositional Language in Multi-Agent Populations](https://shagunsodhani.in/papers-I-read/Emergence-of-Grounded-Compositional-Language-in-Multi-Agent-Populations)
 * [A Semantic Loss Function for Deep Learning with Symbolic Knowledge](https://shagunsodhani.in/papers-I-read/A-Semantic-Loss-Function-for-Deep-Learning-with-Symbolic-Knowledge)
 * [Hierarchical Graph Representation Learning with Differentiable Pooling](https://shagunsodhani.in/papers-I-read/Hierarchical-Graph-Representation-Learning-with-Differentiable-Pooling)

site/_posts/2018-09-27-HoME - a Household Multimodal Environment.md

@@ -0,0 +1,64 @@
+---
+layout: post
+title: HoME - a Household Multimodal Environment
+comments: True
+excerpt: 
+tags: ['2017', 'Multi Modal', 'NIPS 2017', 'NIPS Workskop', 'Virtual Embodiment', AI, NIPS]
+---
+
+## Introduction
+
+* Environment for learning using modalities like vision, audio, semantics, physics and interaction with objects and other agents.
+
+* [Link to the paper](https://arxiv.org/abs/1711.11017)
+
+## Motivation
+
+* Humans learn by interacting with their surroundings (environment).
+
+* Similarly training an agent in an interactive multi-model environment (virtual embodiment) could be useful for a learning agent.
+
+
+## Characteristics
+
+* Open-source and Open-AI gym compatible
+
+* Built on top of 45000 3D house layouts from SUNCG dataset.
+
+* Provides both 3D visual and audio recording.
+
+* Semantic image segmentation and langauge description of objects.
+
+## Components
+
+* Rendering Engine
+
+    * Implemented using Panda 3D game engine.
+
+    * Renders RGB+depth scenes based on textures, multi-source lightings and shadows.
+
+* Acoustic Engine
+
+    * Implemented using EVERT
+
+    * Supports multiple microphones, sound sources, sound absorption based on material, atmospheric conditions etc.
+
+* Semantics Engine
+
+    * Provides a short textual description for each object, along with information like color, category, material size, location etc.
+
+* Physics Engine
+
+    * Implemented using Bullet3 Engine
+
+    * Supports physical interaction, external forces like gravity and position and velocity information for multiple agents.
+
+## Potential Applications
+
+* Visual Question Answering
+
+* Conversational Agents
+
+* Training an agent to follow instructions
+
+* Multi-agent communication