Group equivariant convolutional neural networks (G-CNNs) have recently emerged as a very effective model class for learning from signals in the context of known symmetries. A wide variety of equivariant layers has been proposed for signals on 2D and 3D Euclidean space, graphs, and the sphere, and it has become difficult to see how all of these methods are related, and how they may be generalized. In this paper, we present a fairly general theory of equivariant convolutional networks. Convolutional feature spaces are described as fields over a homogeneous base space, such as the plane $\mathbb{R}^2$, sphere $S^2$ or a graph $\mathcal{G}$. The theory enables a systematic classification of all existing G-CNNs in terms of their group of symmetry, base space, and field type (e.g. scalar, vector, or tensor field, etc.). In addition to this classification, we use Mackey theory to show that convolutions with equivariant kernels are the most general class of equivariant maps between such fields, thus establishing G-CNNs as a universal class of equivariant networks. The theory also explains how the space of equivariant kernels can be parameterized for learning, thereby simplifying the development of G-CNNs for new spaces and symmetries. Finally, the theory introduces a rich geometric semantics to learned feature spaces, thus improving interpretability of deep networks, and establishing a connection to central ideas in mathematics and physics.

Classification tasks are usually analysed and improved through new model architectures or hyperparameter optimisation but the underlying properties of datasets are discovered on an ad-hoc basis as errors occur. However, understanding the properties of the data is crucial in perfecting models. In this paper we analyse exactly which characteristics of a dataset best determine how difficult that dataset is for the task of text classification. We then propose an intuitive measure of difficulty for text classification datasets which is simple and fast to calculate. We show that this measure generalises to unseen data by comparing it to state-of-the-art datasets and results. This measure can be used to analyse the precise source of errors in a dataset and allows fast estimation of how difficult a dataset is to learn. We searched for this measure by training 12 classical and neural network based models on 78 real-world datasets, then use a genetic algorithm to discover the best measure of difficulty. Our difficulty-calculating code ( https://github.com/Wluper/edm ) and datasets ( http://data.wluper.com ) are publicly available.

Humans have consciousness as the ability to perceive events and objects: a mental model of the world developed from the most impoverished of visual stimuli, enabling humans to make rapid decisions and take actions. Although spatial and temporal aspects of different scenes are generally diverse, the underlying physics among environments still work the same way, thus learning an abstract description of shared physical dynamics helps human to understand the world. In this paper, we explore building this mental world with neural network models through multi-task learning, namely the meta-world model. We show through extensive experiments that our proposed meta-world models successfully capture the common dynamics over the compact representations of visually different environments from Atari Games. We also demonstrate that agents equipped with our meta-world model possess the ability of visual self-recognition, i.e., recognize themselves from the reflected mirrored environment derived from the classic mirror self-recognition test (MSR).

A wareness plays a major role in human cognition and adaptive behaviour, though mechanisms involved remain unknown. A wareness is not an objectively established fact, therefore, despite extensive research, scientists have not been able to fully interpret its contribution in multisensory integration and precise neural firing, hence, questions remain: (1) How the biological neuron integrates the incoming multisensory signals with respect to different situations? Recently, scientists have exploited deep learning architectures to integrate multimodal cues and capture context-dependent meanings. Y et, these methods suffer from imprecise behavioural representation and a limited understanding of neural circuitry or underlying information processing mechanisms with respect to the outside world. In this research, we introduce a new theory on the role of awareness and universal context that can help answering the aforementioned crucial neuroscience questions. Specifically, we propose a class of spiking conscious neuron in which the output depends on three functionally distinctive integrated input variables: receptive field (RF), local contextual field (LCF), and universal contextual field (UCF) - a newly proposed dimension. The RF defines the incoming ambiguous sensory signal, LCF defines the modulatory sensory signal coming from other parts of the brain, and UCF defines the awareness. It is believed that the conscious neuron inherently contains enough knowledge about the situation in which the problem is to be solved based on past learning and reasoning and it defines the precise role of incoming multisensory signals (amplification or attenuation) to originate a precise neural firing (exhibiting switch-like behaviour). It is shown, when implemented within an SCNN, the conscious neuron helps modelling a more precise human behaviour e.g., when exploited to model human audiovisual speech processing, the SCNN performed comparably to deep long-short-term memory (LSTM) network. We believe that the proposed theory could be applied to address a range of real-world problems including elusive neural disruptions, explainable artificial intelligence, humanlike computing, low-power neuromorphic chips etc. Keywords: Multisensory Integration, Conscious Neuron, Behavioural Modelling1.

In this work, a non-gradient descent learning scheme is proposed for deep feedforward neural networks (DNN). As we known, autoencoder can be used as the building blocks of the multi-layer perceptron (MLP) deep neural network. So, the MLP will be taken as an example to illustrate the proposed scheme of pseudoinverse learning algorithm for autoencoder (PILAE) training. The PILAE with low rank approximation is a non-gradient based learning algorithm, and the encoder weight matrix is set to be the low rank approximation of the pseudoinverse of the input matrix, while the decoder weight matrix is calculated by the pseudoinverse learning algorithm. It is worth to note that only few network structure hyperparameters need to be tuned. Hence, the proposed algorithm can be regarded as a quasi-automated training algorithm which can be utilized in autonomous machine learning research field. The experimental results show that the proposed learning scheme for DNN can achieve better performance on considering the tradeoff between training efficiency and classification accuracy.

Transfer learning for deep neural networks is the process of first training a base network on a source dataset, and then transferring the learned features (the network's weights) to a second network to be trained on a target dataset. This idea has been shown to improve deep neural network's generalization capabilities in many computer vision tasks such as image recognition and object localization. Apart from these applications, deep Convolutional Neural Networks (CNNs) have also recently gained popularity in the Time Series Classification (TSC) community. However, unlike for image recognition problems, transfer learning techniques have not yet been investigated thoroughly for the TSC task. This is surprising as the accuracy of deep learning models for TSC could potentially be improved if the model is fine-tuned from a pre-trained neural network instead of training it from scratch. In this paper, we fill this gap by investigating how to transfer deep CNNs for the TSC task. To evaluate the potential of transfer learning, we performed extensive experiments using the UCR archive which is the largest publicly available TSC benchmark containing 85 datasets. For each dataset in the archive, we pre-trained a model and then fine-tuned it on the other datasets resulting in 7140 different deep neural networks. These experiments revealed that transfer learning can improve or degrade the model's predictions depending on the dataset used for transfer. Therefore, in an effort to predict the best source dataset for a given target dataset, we propose a new method relying on Dynamic Time Warping to measure inter-datasets similarities. We describe how our method can guide the transfer to choose the best source dataset leading to an improvement in accuracy on 71 out of 85 datasets.

We present a simple and general framework for feature learning from point clouds. The key to the success of CNNs is the convolution operator that is capable of leveraging spatially-local correlation in data represented densely in grids (e.g. images). However, point clouds are irregular and unordered, thus directly convolving kernels against features associated with the points, will result in desertion of shape information and variance to point ordering. To address these problems, we propose to learn an $\mathcal{X}$-transformation from the input points, to simultaneously promote two causes. The first is the weighting of the input features associated with the points, and the second is the permutation of the points into a latent and potentially canonical order. Element-wise product and sum operations of the typical convolution operator are subsequently applied on the $\mathcal{X}$-transformed features. The proposed method is a generalization of typical CNNs to feature learning from point clouds, thus we call it PointCNN. Experiments show that PointCNN achieves on par or better performance than state-of-the-art methods on multiple challenging benchmark datasets and tasks.

Data scientists are expected to know a lot -- machine learning, computer science, statistics, mathematics, data visualization, communication, and deep learning. How should data scientists who want to be in demand by employers spend their learning budget? I scoured job listing websites to find which skills are most in demand for data scientists. I looked at general data science skills and at specific languages and tools separately. I searched job listings on LinkedIn, Indeed, SimplyHired, Monster, and AngelList on October 10, 2018.

Not too long ago it was often said that computer vision could not compete with the visual abilities of a one-year-old. That is no longer true: computers can now recognize objects in images about as well as most adults can, and there are computerized cars on the road that drive themselves more safely than an average sixteen-year-old could. And rather than being told how to see or drive, computers have learned from experience, following a path that nature took millions of years ago. What is fueling these advances is gushers of data. Data are the new oil. Learning algorithms are refineries that extract information from raw data; information can be used to create knowledge; knowledge leads to understanding; and understanding leads to wisdom. Welcome to the brave new world of deep learning. Deep learning is a branch of machine learning that has its roots in mathematics, computer science, and neuroscience.

Depending on what news headline you have read, you may have perceived an Artificial Intelligence (AI) system as either an Alexa or Siri assistant that understands all your commands, a deep learning system that can recognize dog or a cat from image, a system that recommends personalized medicine, or an intelligent, overpowering machine that can overtake all human tasks and render humans useless. Few of these definitions can be termed as visionary, few fear mongering and rest of them being evolutionary. Last month, I was at Artificial Intelligence (AI) Summit 2018 in San Francisco. The event highlighted the state of AI business as it stands today and real-world examples from enterprises who are using AI to transform their business. I want to give a shout out to organizers of AI Summit – they did a fabulous job in bringing a highly diverse set of speakers across a variety of verticals.