Most popular optimizers for deep learning can be broadly categorized as adaptive methods (e.g. Adam) and accelerated schemes (e.g. For many models such as convolutional neural networks (CNNs), adaptive methods typically converge faster but generalize worse compared to SGD; for complex settings such as generative adversarial networks (GANs), adaptive methods are typically the default because of their stability. We propose AdaBelief to simultaneously achieve three goals: fast convergence as in adaptive methods, good generalization as in SGD, and training stability. The intuition for AdaBelief is to adapt the stepsize according to the "belief" in the current gradient direction.

We developed a machine learning model which is capable of completing an outfit based on a given seed product. Here we give an overview of our model and some of the challenges we faced. We consider an outfit to be a set of fashion items which match stylistically and can be worn together. In order for the outfit to work, each item must be compatible with all other items. Our aim is to create a model which embeds each item in a latent style space such that for any two items the dot product (a measure of similarity) of their embeddings reflects their compatibility.

Intelligently reasoning about the world often requires integrating data from multiple modalities, as any individual modality may contain unreliable or incomplete information. On the other hand, the brain performs multimodal processing almost immediately. This divide between conventional multimodal learning and neuroscience suggests that a detailed study of early multimodal fusion could improve artificial multimodal representations. To facilitate the study of early multimodal fusion, we create a convolutional LSTM network architecture that simultaneously processes both audio and visual inputs, and allows us to select the layer at which audio and visual information combines. Our results demonstrate that immediate fusion of audio and visual inputs in the initial C-LSTM layer results in higher performing networks that are more robust to the addition of white noise in both audio and visual inputs. In many cases, an individual modality does not contain sufficient information to classify the scene.

While humans are aware of their body and capabilities, robots are not. To address this, we present in this paper a neural network architecture that enables a dual-arm robot to get a sense of itself in an environment. Our approach is inspired by human self-awareness developmental levels and serves as the underlying building block for a robot to achieve awareness of itself while carrying out tasks in an environment. We assume that a robot has to know itself before interacting with the environment in order to be able to support different robotic tasks. Hence, we implemented a neural network architecture to enable a robot to differentiate its limbs from the environment using visual and proprioception sensory inputs. We demonstrate experimentally that a robot can distinguish itself with an accuracy of 88.7% on average in cluttered environmental settings and under confounding input signals.

One of the main issues for underwater robots navigation is their accurate positioning, which heavily depends on the orientation estimation phase. The systems employed to this scope are affected by different noise typologies, mainly related to the sensors and to the irregular noise of the underwater environment. Filtering algorithms can reduce their effect if opportunely configured, but this process usually requires fine techniques and time. In this paper we propose DANAE, a deep Denoising AutoeNcoder for Attitude Estimation which works on Kalman filter IMU/AHRS data integration with the aim of reducing any kind of noise, independently of its nature. This deep learning-based architecture showed to be robust and reliable, significantly improving the Kalman filter results. Further tests could make this method suitable for real-time applications on navigation tasks.

Millimeter-wave (mmW) radars are being increasingly integrated into commercial vehicles to support new advanced driver-assistance systems (ADAS) by enabling robust and high-performance object detection, localization, as well as recognition - a key component of new environmental perception. In this paper, we propose a novel radar multiple-perspectives convolutional neural network (RAMP-CNN) that extracts the location and class of objects based on further processing of the range-velocity-angle (RVA) heatmap sequences. To bypass the complexity of 4D convolutional neural networks (NN), we propose to combine several lower-dimension NN models within our RAMP-CNN model that nonetheless approaches the performance upper-bound with lower complexity. The extensive experiments show that the proposed RAMP-CNN model achieves better average recall (AR) and average precision (AP) than prior works in all testing scenarios (see Table. III). Besides, the RAMP-CNN model is validated to work robustly under the nighttime, which enables low-cost radars as a potential substitute for pure optical sensing under severe conditions.

With data durability, high access speed, low power efficiency and byte addressability, NVMe and SSD, which are acknowledged representatives of emerging storage technologies, have been applied broadly in many areas. However, one key issue with high-performance adoption of these technologies is how to properly define intelligent cache layers such that the performance gap between emerging technologies and main memory can be well bridged. To this end, we propose Phoebe, a reuse-aware reinforcement learning framework for the optimal online caching that is applicable for a wide range of emerging storage models. By continuous interacting with the cache environment and the data stream, Phoebe is capable to extract critical temporal data dependency and relative positional information from a single trace, becoming ever smarter over time. To reduce training overhead during online learning, we utilize periodical training to amortize costs. Phoebe is evaluated on a set of Microsoft cloud storage workloads. Experiment results show that Phoebe is able to close the gap of cache miss rate from LRU and a state-of-the-art online learning based cache policy to the Belady's optimal policy by 70.3% and 52.6%, respectively.

Convex optimizers have known many applications as differentiable layers within deep neural architectures. One application of these convex layers is to project points into a convex set. However, both forward and backward passes of these convex layers are significantly more expensive to compute than those of a typical neural network. We investigate in this paper whether an inexact, but cheaper projection, can drive a descent algorithm to an optimum. Specifically, we propose an interpolation-based projection that is computationally cheap and easy to compute given a convex, domain defining, function. We then propose an optimization algorithm that follows the gradient of the composition of the objective and the projection and prove its convergence for linear objectives and arbitrary convex and Lipschitz domain defining inequality constraints. In addition to the theoretical contributions, we demonstrate empirically the practical interest of the interpolation projection when used in conjunction with neural networks in a reinforcement learning and a supervised learning setting.

The recently introduced locally orderless tensor network (LoTeNet) for supervised image classification uses matrix product state (MPS) operations on grids of transformed image patches. The resulting patch representations are combined back together into the image space and aggregated hierarchically using multiple MPS blocks per layer to obtain the final decision rules. In this work, we propose a non-patch based modification to LoTeNet that performs one MPS operation per layer, instead of several patch-level operations. The spatial information in the input images to MPS blocks at each layer is squeezed into the feature dimension, similar to LoTeNet, to maximise retained spatial correlation between pixels when images are flattened into 1D vectors. The proposed multi-layered tensor network (MLTN) is capable of learning linear decision boundaries in high dimensional spaces in a multi-layered setting, which results in a reduction in the computation cost compared to LoTeNet without any degradation in performance.

Deep learning applied to weather forecasting has started gaining popularity because of the progress achieved by data-driven models. The present paper compares four different deep learning architectures to perform weather prediction on daily data gathered from 18 cities across Europe and spanned over a period of 15 years. The four proposed models investigate the different type of input representations (i.e. tensorial unistream vs. multi-stream matrices) as well as the combination of convolutional neural networks and LSTM (i.e. cascaded vs. ConvLSTM). In particular, we show that a model that uses a multi-stream input representation and that processes each lag individually combined with a cascaded convolution and LSTM is capable of better forecasting than the other compared models. In addition, we show that visualization techniques such as occlusion analysis and score maximization can give an additional insight on the most important features and cities for predicting a particular target feature and city.