to

### Stochastic Pooling for Regularization of Deep Convolutional Neural Networks

We introduce a simple and effective method for regularizing large convolutional neural networks. We replace the conventional deterministic pooling operations with a stochastic procedure, randomly picking the activation within each pooling region according to a multinomial distribution, given by the activities within the pooling region. The approach is hyper-parameter free and can be combined with other regularization approaches, such as dropout and data augmentation. We achieve state-of-the-art performance on four image datasets, relative to other approaches that do not utilize data augmentation.

### 2D-3D Geometric Fusion Network using Multi-Neighbourhood Graph Convolution for RGB-D Indoor Scene Classification

Multi-modal fusion has been proved to help enhance the performance of scene classification tasks. This paper presents a 2D-3D fusion stage that combines 3D Geometric features with 2D Texture features obtained by 2D Convolutional Neural Networks. To get a robust 3D Geometric embedding, a network that uses two novel layers is proposed. The first layer, Multi-Neighbourhood Graph Convolution, aims to learn a more robust geometric descriptor of the scene combining two different neighbourhoods: one in the Euclidean space and the other in the Feature space. The second proposed layer, Nearest Voxel Pooling, improves the performance of the well-known Voxel Pooling. Experimental results, using NYU-Depth-v2 and SUN RGB-D datasets, show that the proposed method outperforms the current state-of-the-art in RGB-D indoor scene classification tasks.

### Grab Halts Late-Night Car-Pooling in Singapore After Driver Complaints

Citing feedback from drivers who described intoxicated commuters vomiting in their cars or arguing with them during the ride, the company said it would stop its GrabHitch service between 1 a.m. and 5 a.m. from next month.

### Partial Pooling for Lower Variance Variable Encoding

We can generate a synthetic data set according to these assumptions, with distributions similar to the distributions observed in the radon data set that we used in our earlier post: 85 groups, sampled unevenly. Here, we take a peek at our data, df. As the graph shows, some groups were heavily sampled, but most groups have only a handful of samples in the data set. Since this is synthetic data, we know the true population means (shown in red in the graph below), and we can compare them to the observed means $$\bar{y}_i$$ of each group $$i$$ (shown in black, with standard errors. The gray points are the actual observations).

### Didi redesigns car-pooling service in China after death of female passenger

Chinese car-hailing platform Didi Chuxing has quietly resumed its car-pooling services at the weekend, with major functions undergoing a redesign, aimed at ensuring the safety of passengers. The service has been halted since May 12 after a 21-year-old flight attendant reportedly died at the hands of a 26-year-old driver in China's Henan province on May 6. The flight attendant had booked a ride through Didi and was allegedly killed by the driver who was also found dead in a river by police, according to Chinese news reports. There are some things that machines are simply better at doing than humans, but humans still have plenty going for them. Here's a look at how the two are going to work in concert to deliver a more powerful future for IT, and the human race.