In this paper, we aim to understand Residual Network (ResNet) in a scientifically sound way by providing a bridge between ResNet and Feynman path integral. In particular, we prove that the effect of residual block is equivalent to partial differential equation, and the ResNet transforming process can be equivalently converted to Feynman path integral. These conclusions greatly help us mathematically understand the advantage of ResNet in addressing the gradient vanishing issue. More importantly, our analyses offer a path integral view of ResNet, and demonstrate that the output of certain network can be obtained by adding contributions of all paths. Moreover, the contribution of each path is proportional to e^{-S}, where S is the action given by time integral of Lagrangian L. This lays the solid foundation in the understanding of ResNet, and provides insights in the future design of convolutional neural network architecture. Based on these results, we have designed the network using partial differential operators, which further validates our theoritical analyses.

Object detection in still images has drawn a lot of attention over past few years, and with the advent of Deep Learning impressive performances have been achieved with numerous industrial applications. Most of these deep learning models rely on RGB images to localize and identify objects in the image. However in some application scenarii, images are compressed either for storage savings or fast transmission. Therefore a time consuming image decompression step is compulsory in order to apply the aforementioned deep models. To alleviate this drawback, we propose a fast deep architecture for object detection in JPEG images, one of the most widespread compression format. We train a neural network to detect objects based on the blockwise DCT (discrete cosine transform) coefficients {issued from} the JPEG compression algorithm. We modify the well-known Single Shot multibox Detector (SSD) by replacing its first layers with one convolutional layer dedicated to process the DCT inputs. Experimental evaluations on PASCAL VOC and industrial dataset comprising images of road traffic surveillance show that the model is about $2\times$ faster than regular SSD with promising detection performances. To the best of our knowledge, this paper is the first to address detection in compressed JPEG images.

Is cognition a collection of loosely connected functions tuned to different tasks, or can there be a general learning algorithm? If such an hypothetical general algorithm did exist, tuned to our world, could it adapt seamlessly to a world with different laws of nature? We consider the theory that predictive coding is such a general rule, and falsify it for one specific neural architecture known for high-performance predictions on natural videos and replication of human visual illusions: PredNet. Our results show that PredNet's high performance generalizes without retraining on a completely different natural video dataset. Yet PredNet cannot be trained to reach even mediocre accuracy on an artificial video dataset created with the rules of the Game of Life (GoL). We also find that a submodule of PredNet, a Convolutional Neural Network trained alone, reaches perfect accuracy on the GoL while being mediocre for natural videos, showing that PredNet's architecture itself is responsible for both the high performance on natural videos and the loss of performance on the GoL. Just as humans cannot predict the dynamics of the GoL, our results suggest that there might be a trade-off between high performance on sensory inputs with different sets of rules.

Object-based approaches for learning action-conditioned dynamics has demonstrated promise for generalization and interpretability. However, existing approaches suffer from structural limitations and optimization difficulties for common environments with multiple dynamic objects. In this paper, we present a novel self-supervised learning framework, called Multi-level Abstraction Object-oriented Predictor (MAOP), which employs a three-level learning architecture that enables efficient object-based dynamics learning from raw visual observations. We also design a spatial-temporal relational reasoning mechanism for MAOP to support instance-level dynamics learning and handle partial observability. Our results show that MAOP significantly outperforms previous methods in terms of sample efficiency and generalization over novel environments for learning environment models. We also demonstrate that learned dynamics models enable efficient planning in unseen environments, comparable to true environment models. In addition, MAOP learns semantically and visually interpretable disentangled representations.

Cross-view image translation is challenging because it involves images with drastically different views and severe deformation. In this paper, we propose a novel approach named Multi-Channel Attention SelectionGAN (SelectionGAN) that makes it possible to generate images of natural scenes in arbitrary viewpoints, based on an image of the scene and a novel semantic map. The proposed SelectionGAN explicitly utilizes the semantic information and consists of two stages. In the first stage, the condition image and the target semantic map are fed into a cycled semantic-guided generation network to produce initial coarse results. In the second stage, we refine the initial results by using a multi-channel attention selection mechanism. Moreover, uncertainty maps automatically learned from attentions are used to guide the pixel loss for better network optimization. Extensive experiments on Dayton, CVUSA and Ego2Top datasets show that our model is able to generate significantly better results than the state-of-the-art methods. The source code, data and trained models are available at https://github.com/Ha0Tang/SelectionGAN.

Diabetes is a globally prevalent disease that can cause visible microvascular complications such as diabetic retinopathy and macular edema in the human eye retina, the images of which are today used for manual disease screening. This labor-intensive task could greatly benefit from automatic detection using deep learning technique. Here we present a deep learning system that identifies referable diabetic retinopathy comparably or better than presented in the previous studies, although we use only a small fraction of images ( 1/4) in training but are aided with higher image resolutions. We also provide novel results for five different screening and clinical grading systems for diabetic retinopathy and macular edema classification, including results for accurately classifying images according to clinical five-grade diabetic retinopathy and four-grade diabetic macular edema scales. These results suggest, that a deep learning system could increase the cost-effectiveness of screening while attaining higher than recommended performance, and that the system could be applied in clinical examinations requiring finer grading.

In this paper we propose an end-to-end LSTM-based model that performs single-channel speech enhancement and phone recognition in a cocktail party scenario where visual information of the target speaker is available. In the speech enhancement phase the proposed system uses a "visual attention" signal of the speaker of interest to extract her speech from the input mixed-speech signal, while in the ASR phase it recognizes her phone sequence through a phone recognizer trained with a CTC loss. It is well known that learning multiple related tasks from data simultaneously can improve performance than learning these tasks independently, therefore we decided to train the model by optimizing both tasks at the same time. This allowed us also to explore whether (and how) this joint optimization leads to better results. We analyzed different training strategies that reveal some interesting and unexpected behaviors. In particular, the experiments demonstrated that during optimization of the ASR phase the speech enhancement capability of the model significantly decreases and vice-versa. We evaluated our approach on mixed-speech versions of GRID and TCD-TIMIT. The obtained results show a remarkable drop of the Phone Error Rate (PER) compared to the audio-visual baseline models trained only to perform phone recognition phase.

Dropout is commonly used to help reduce overfitting in deep neural networks. Sparsity is a potentially important property of neural networks, but is not explicitly controlled by Dropout-based regularization. In this work, we propose Sparseout a simple and efficient variant of Dropout that can be used to control the sparsity of the activations in a neural network. We theoretically prove that Sparseout is equivalent to an $L_q$ penalty on the features of a generalized linear model and that Dropout is a special case of Sparseout for neural networks. We empirically demonstrate that Sparseout is computationally inexpensive and is able to control the desired level of sparsity in the activations. We evaluated Sparseout on image classification and language modelling tasks to see the effect of sparsity on these tasks. We found that sparsity of the activations is favorable for language modelling performance while image classification benefits from denser activations. Sparseout provides a way to investigate sparsity in state-of-the-art deep learning models. Source code for Sparseout could be found at \url{https://github.com/najeebkhan/sparseout}.

Multi-label classification (MLC) is the task of assigning a set of target labels for a given sample. Modeling the combinatorial label interactions in MLC has been a long-haul challenge. We propose Label Message Passing (LaMP) Neural Networks to efficiently model the joint prediction of multiple labels. LaMP treats labels as nodes on a label-interaction graph and computes the hidden representation of each label node conditioned on the input using attention-based neural message passing. Attention enables LaMP to assign different importance to neighbor nodes per label, learning how labels interact (implicitly). The proposed models are simple, accurate, interpretable, structure-agnostic, and applicable for predicting dense labels since LaMP is incredibly parallelizable. We validate the benefits of LaMP on seven real-world MLC datasets, covering a broad spectrum of input/output types and outperforming the state-of-the-art results. Notably, LaMP enables intuitive interpretation of how classifying each label depends on the elements of a sample and at the same time rely on its interaction with other labels. We provide our code and datasets at https://github.com/QData/LaMP

Machine learning has made major advances in categorizing objects in images, yet the best algorithms miss important aspects of how people learn and think about categories. People can learn richer concepts from fewer examples, including causal models that explain how members of a category are formed. Here, we explore the limits of this human ability to infer causal "programs" -- latent generating processes with nontrivial algorithmic properties -- from one, two, or three visual examples. People were asked to extrapolate the programs in several ways, for both classifying and generating new examples. As a theory of these inductive abilities, we present a Bayesian program learning model that searches the space of programs for the best explanation of the observations. Although variable, people's judgments are broadly consistent with the model and inconsistent with several alternatives, including a pre-trained deep neural network for object recognition, indicating that people can learn and reason with rich algorithmic abstractions from sparse input data.