Neural Architecture Search (NAS) aims to facilitate the design of deep networks for new tasks. Existing techniques rely on two stages: searching over the architecture space and validating the best architecture. Evaluating NAS algorithms is currently solely done by comparing their results on the downstream task. While intuitive, this fails to explicitly evaluate the effectiveness of their search strategies. In this paper, we extend the NAS evaluation procedure to include the search phase. To this end, we compare the quality of the solutions obtained by NAS search policies with that of random architecture selection. We find that: (i) On average, the random policy outperforms state-of-the-art NAS algorithms; and (ii) The results and candidate rankings of NAS algorithms do not reflect the true performance of the candidate architectures. While our former finding illustrates the fact that the NAS search space has been sufficiently constrained so that random solutions yield good results, we trace the latter back to the weight sharing strategy used by state-of-the-art NAS methods. In contrast with common belief, weight sharing negatively impacts the training of good architectures, thus reducing the effectiveness of the search process. We believe that following our evaluation framework will be key to designing NAS strategies that truly discover superior architectures.

The softmax function on top of a final linear layer is the de facto method to output probability distributions in neural networks. In many applications such as language models or text generation, this model has to produce distributions over large output vocabularies. Recently, this has been shown to have limited representational capacity due to its connection with the rank bottleneck in matrix factorization. However, little is known about the limitations of linear-softmax for quantities of practical interest such as cross entropy or mode estimation, a direction that we theoretically and empirically explore here. As an efficient and effective solution to alleviate this issue, we propose to learn parametric monotonic functions on top of the logits. We theoretically investigate the rank increasing capabilities of such monotonic functions. Empirically, our method improves in two different quality metrics over the traditional softmax-linear layer in synthetic and real language model experiments, adding little time or memory overhead, while being comparable to the more computationally expensive mixture of softmaxes.

Perinatal stroke (PS) is a serious condition that, if undetected and thus untreated, often leads to life-long disability, in particular Cerebral Palsy (CP). In clinical settings, Prechtl's General Movement Assessment (GMA) can be used to classify infant movements using a Gestalt approach, identifying infants at high risk of developing PS. Training and maintenance of assessment skills are essential and expensive for the correct use of GMA, yet many practitioners lack these skills, preventing larger-scale screening and leading to significant risks of missing opportunities for early detection and intervention for affected infants. We present an automated approach to GMA, based on body-worn accelerometers and a novel sensor data analysis method-Discriminative Pattern Discovery (DPD)-that is designed to cope with scenarios where only coarse annotations of data are available for model training. We demonstrate the effectiveness of our approach in a study with 34 newborns (21 typically developing infants and 13 PS infants with abnormal movements). Our method is able to correctly recognise the trials with abnormal movements with at least the accuracy that is required by newly trained human annotators (75%), which is encouraging towards our ultimate goal of an automated PS screening system that can be used population-wide.

We explore the use of graph networks to deal with irregular-geometry detectors in the context of particle reconstruction. Thanks to their representation-learning capabilities, graph networks can exploit the full detector granularity, while natively managing the event sparsity and arbitrarily complex detector geometries. We introduce two distance-weighted graph network architectures, dubbed GarNet and GravNet layers, and apply them to a typical particle reconstruction task. The performance of the new architectures is evaluated on a data set of simulated particle interactions on a toy model of a highly granular calorimeter, loosely inspired by the endcap calorimeter to be installed in the CMS detector for the High-Luminosity LHC phase. We study the clustering of energy depositions, which is the basis for calorimetric particle reconstruction, and provide a quantitative comparison to alternative approaches. The proposed algorithms outperform existing methods or reach competitive performance with lower computing-resource consumption. Being geometry-agnostic, the new architectures are not restricted to calorimetry and can be easily adapted to other use cases, such as tracking in silicon detectors.

Dimensionality reduction methods, also known as projections, are frequently used for exploring multidimensional data in machine learning, data science, and information visualization. Among these, t-SNE and its variants have become very popular for their ability to visually separate distinct data clusters. However, such methods are computationally expensive for large datasets, suffer from stability problems, and cannot directly handle out-of-sample data. We propose a learning approach to construct such projections. We train a deep neural network based on a collection of samples from a given data universe, and their corresponding projections, and next use the network to infer projections of data from the same, or similar, universes. Our approach generates projections with similar characteristics as the learned ones, is computationally two to three orders of magnitude faster than SNE-class methods, has no complex-to-set user parameters, handles out-of-sample data in a stable manner, and can be used to learn any projection technique. We demonstrate our proposal on several real-world high dimensional datasets from machine learning.

Inter-Cell Interference Coordination (ICIC) is a promising way to improve energy efficiency in wireless networks, especially where small base stations are densely deployed. However, traditional optimization based ICIC schemes suffer from severe performance degradation with complex interference pattern. To address this issue, we propose a Deep Reinforcement Learning with Deterministic Policy and Target (DRL-DPT) framework for ICIC in wireless networks. DRL-DPT overcomes the main obstacles in applying reinforcement learning and deep learning in wireless networks, i.e. continuous state space, continuous action space and convergence. Firstly, a Deep Neural Network (DNN) is involved as the actor to obtain deterministic power control actions in continuous space. Then, to guarantee the convergence, an online training process is presented, which makes use of a dedicated reward function as the target rule and a policy gradient descent algorithm to adjust DNN weights. Experimental results show that the proposed DRL-DPT framework consistently outperforms existing schemes in terms of energy efficiency and throughput under different wireless interference scenarios. More specifically, it improves up to 15% of energy efficiency with faster convergence rate.

As humans we are driven by a strong desire for seeking novelty in our world. Also upon observing a novel pattern we are capable of refining our understanding of the world based on the new information---humans can discover their world. The outstanding ability of the human mind for discovery has led to many breakthroughs in science, art and technology. Here we investigate the possibility of building an agent capable of discovering its world using the modern AI technology. In particular we introduce NDIGO, Neural Differential Information Gain Optimisation, a self-supervised discovery model that aims at seeking new information to construct a global view of its world from partial and noisy observations. Our experiments on some controlled 2-D navigation tasks show that NDIGO outperforms state-of-the-art information-seeking methods in terms of the quality of the learned representation. The improvement in performance is particularly significant in the presence of white or structured noise where other information-seeking methods follow the noise instead of discovering their world.

While many approaches to make neural networks more fathomable have been proposed, they are restricted to interrogating the network with input data. Measures for characterizing and monitoring structural properties, however, have not been developed. In this work, we propose neural persistence, a complexity measure for neural network architectures based on topological data analysis on weighted stratified graphs. To demonstrate the usefulness of our approach, we show that neural persistence reflects best practices developed in the deep learning community such as dropout and batch normalization. Moreover, we derive a neural persistence-based stopping criterion that shortens the training process while achieving comparable accuracies as early stopping based on validation loss.

End-to-end optimization has achieved state-of-the-art performance on many specific problems, but there is no straight-forward way to combine pretrained models for new problems. Here, we explore improving modularity by learning a post-hoc interface between two existing models to solve a new task. Specifically, we take inspiration from neural machine translation, and cast the challenging problem of cross-modal domain transfer as unsupervised translation between the latent spaces of pretrained deep generative models. By abstracting away the data representation, we demonstrate that it is possible to transfer across different modalities (e.g., image-to-audio) and even different types of generative models (e.g., VAE-to-GAN). We compare to state-of-the-art techniques and find that a straight-forward variational autoencoder is able to best bridge the two generative models through learning a shared latent space. We can further impose supervised alignment of attributes in both domains with a classifier in the shared latent space. Through qualitative and quantitative evaluations, we demonstrate that locality and semantic alignment are preserved through the transfer process, as indicated by high transfer accuracies and smooth interpolations within a class. Finally, we show this modular structure speeds up training of new interface models by several orders of magnitude by decoupling it from expensive retraining of base generative models.

Spatio-temporal data are ubiquitous in the agricultural, ecological, and environmental sciences, and their study is important for understanding and predicting a wide variety of processes. One of the difficulties with modeling spatial processes that change in time is the complexity of the dependence structures that must describe how such a process varies, and the presence of high-dimensional complex data sets and large prediction domains. It is particularly challenging to specify parameterizations for nonlinear dynamic spatio-temporal models (DSTMs) that are simultaneously useful scientifically and efficient computationally. Statisticians have developed deep hierarchical models that can accommodate process complexity as well as the uncertainties in the predictions and inference. However, these models can be expensive and are typically application specific. On the other hand, the machine learning community has developed alternative "deep learning" approaches for nonlinear spatio-temporal modeling. These models are flexible yet are typically not implemented in a probabilistic framework. The two paradigms have many things in common and suggest hybrid approaches that can benefit from elements of each framework. This overview paper presents a brief introduction to the deep hierarchical DSTM (DH-DSTM) framework, and deep models in machine learning, culminating with the deep neural DSTM (DN-DSTM). Recent approaches that combine elements from DH-DSTMs and echo state network DN-DSTMs are presented as illustrations.