- 【代码】A PyTorch Implementation for Densely Connected Convolutional Networks (DenseNets)
This repository contains a PyTorch implementation of the paper Densely Connected Convolutional Networks. The code is based on the excellent PyTorch example for training ResNet on Imagenet.
The detault setting for this repo is a DenseNet-BC (with bottleneck layers and channel reduction), 100 layers, a growth rate of 12 and batch size 128.
The Official torch implementaion contains further links to implementations in other frameworks.
2.【博客】The Two Paths from Natural Language Processing to Artificial Intelligence
Why isn’t Siri smarter? AI has accelerated in recent years, especially with deep learning, but current chatbots are an embarrassment. Computers still can’t read or converse intelligently. Their deficiency is disappointing because we want to interact with our world using natural language, and we want computers to read all of those documents out there so they can retrieve the best ones, answer our questions, and summarize what is new.
To understand our language, computers need to know our world. They need to be able to answer questions like “Why does it only rain outside?” and “If a book is on a table, and you push the table, what happens?”
3.【博客】DeepMind just published a mind blowing paper: PathNet.
Potentially describing how general artificial intelligence will look like.
Since scientists started building and training neural networks, there has always been one barrier called transfer learning. Transfer learning is the capability of an AI to learn from different tasks and apply its pre-learned knowledge to a totally new task. It is implicit that with this precedent knowledge, the AI will perform better and train faster than de novo neural networks on the new task.
DeepMind is maybe on the path of solving this with PathNet. PathNet is a network of neural networks, trained using both stochastic gradient descent and a genetic selection method.
4.【论文 & 代码】ScanNet： Richly-annotated 3D Reconstructions of Indoor Scenes
A key requirement for leveraging supervised deep learning methods is the availability of large, labeled datasets. Unfortunately, in the context of RGB-D scene understanding, very little data is available – current datasets cover a small range of scene views and have limited semantic annotations. To address this issue, we introduce ScanNet, an RGB-D video dataset containing 2.5M views in 1513 scenes annotated with 3D camera poses, surface reconstructions, and semantic segmentations. To collect this data, we designed an easy-to-use and scalable RGB-D capture system that includes automated surface reconstruction and crowdsourced semantic annotation.We show that using this data helps achieve state-of-the-art performance on several 3D scene understanding tasks, including 3D object classification, semantic voxel labeling, and CAD model retrieval. The dataset is freely available at http://www.scan-net.org
5.【博客】TRANSFER LEARNING IN TENSORFLOW USING A PRE-TRAINED INCEPTION-RESNET-V2 MODEL
In this guide, we will see how we can perform transfer learning using the official pre-trained model offered by Google, which can be found in TensorFlow’s model library and downloaded here. As I have mentioned in my previous post on creating TFRecord files, one thing that I find really useful in using TensorFlow-slim over other deep learning libraries is the ready access to the best pretrained models offered by Google. This guide will build upon my previous guide on creating TFRecord files and show you how to use the inception-resnet-v2 model released by Google.