Neural Architecture Search (NAS) methods have been successfully applied to image tasks with excellent results. However, NAS methods are often complex and tend to converge to local minima as soon as generated architectures seem to yield good results. In this paper, we propose G-EA, a novel approach for guided evolutionary NAS. The rationale behind G-EA, is to explore the search space by generating and evaluating several architectures in each generation at initialization stage using a zero-proxy estimator, where only the highest-scoring network is trained and kept for the next generation. This evaluation at initialization stage allows continuous extraction of knowledge from the search space without increasing computation, thus allowing the search to be efficiently guided. Moreover, G-EA forces exploitation of the most performant networks by descendant generation while at the same time forcing exploration by parent mutation and by favouring younger architectures to the detriment of older ones. Experimental results demonstrate the effectiveness of the proposed method, showing that G-EA achieves state-of-the-art results in NAS-Bench-201 search space in CIFAR-10, CIFAR-100 and ImageNet16-120, with mean accuracies of 93.98%, 72.12% and 45.94% respectively.

Importance: Lower-resource areas in Africa and Asia face a unique set of healthcare challenges: the dual high burden of communicable and non-communicable diseases; a paucity of highly trained primary healthcare providers in both rural and densely populated urban areas; and a lack of reliable, inexpensive internet connections. Objective: To address these challenges, we designed an artificial intelligence assistant to help primary healthcare providers in lower-resource areas document demographic and medical sign/symptom data and to record and share diagnostic data in real-time with a centralized database. Design: We trained our system using multiple data sets, including US-based electronic medical records (EMRs) and open-source medical literature and developed an adaptive, general medical assistant system based on machine learning algorithms. Main outcomes and Measure: The application collects basic information from patients and provides primary care providers with diagnoses and prescriptions suggestions. The application is unique from existing systems in that it covers a wide range of common diseases, signs, and medication typical in lower-resource countries; the application works with or without an active internet connection. Results: We have built and implemented an adaptive learning system that assists trained primary care professionals by means of an Android smartphone application, which interacts with a central database and collects real-time data. The application has been tested by dozens of primary care providers. Conclusions and Relevance: Our application would provide primary healthcare providers in lower-resource areas with a tool that enables faster and more accurate documentation of medical encounters. This application could be leveraged to automatically populate local or national EMR systems.

We study anomaly detection for the case when the normal class consists of more than one object category. This is an obvious generalization of the standard one-class anomaly detection problem. However, we show that jointly using multiple one-class anomaly detectors to solve this problem yields poorer results as compared to training a single one-class anomaly detector on all normal object categories together. We further develop a new anomaly detector called DeepMAD that learns compact distinguishing features by exploiting the multiple normal objects categories. This algorithm achieves higher AUC values for different datasets compared to two top performing one-class algorithms that either are trained on each normal object category or jointly trained on all normal object categories combined. In addition to theoretical results we present empirical results using the CIFAR-10, fMNIST, CIFAR-100, and a new dataset we developed called RECYCLE.

We present a neural analysis and synthesis (NANSY) framework that can manipulate voice, pitch, and speed of an arbitrary speech signal. Most of the previous works have focused on using information bottleneck to disentangle analysis features for controllable synthesis, which usually results in poor reconstruction quality. We address this issue by proposing a novel training strategy based on information perturbation. The idea is to perturb information in the original input signal (e.g., formant, pitch, and frequency response), thereby letting synthesis networks selectively take essential attributes to reconstruct the input signal. Because NANSY does not need any bottleneck structures, it enjoys both high reconstruction quality and controllability. Furthermore, NANSY does not require any labels associated with speech data such as text and speaker information, but rather uses a new set of analysis features, i.e., wav2vec feature and newly proposed pitch feature, Yingram, which allows for fully self-supervised training. Taking advantage of fully self-supervised training, NANSY can be easily extended to a multilingual setting by simply training it with a multilingual dataset. The experiments show that NANSY can achieve significant improvement in performance in several applications such as zero-shot voice conversion, pitch shift, and time-scale modification.

In a stark white browser tab, Sam -- a young blonde woman with perfectly shaped lips -- asks me for the solution to 2 2. I immediately think of the infamous Star Trek: The Next Generation episode in which a tortured Captain Picard is shown four lights. If he admits there are five lights, the ordeal will stop. I'm at home, staring at the future face of the metaverse and trying valiantly not to think about memes from a TV show known for its exploration of ethics and humanity. Sam isn't a real person -- she's a digital human created by Auckland-based tech company Soul Machines. Designed to have a short conversation with visitors about herself, she runs on a proprietary "digital brain" and studies my expressions via webcam.

An expansive compound of buildings covering the equivalent of 46 football pitches was recently erected on the outskirts of Guangzhou, China's bustling southern metropolis. The sprawling complex of three-storey buildings contains some 5,000 rooms and is the first of what is expected to be a chain of quarantine centres built by the Chinese government to house people arriving from overseas as it forges ahead with its zero-tolerance approach to COVID. The compound is equipped with "5G communication technology and artificial intelligence" infrastructure, and each room, which can host only one person at a time, has cameras at its door and a robot delivery system to "minimise human contact and the risk of cross-infection", according to the introduction to the centre put out by the Guangzhou government. It took the construction team less than three months to finish the project – in an echo of the Huoshenshan and Leishenshan temporary hospitals that were built in record time in the central city of Wuhan as COVID-19 took hold in early 2020. But while those hospitals were greeted with relief, the appearance of the quarantine centre nearly two years after the trauma of Wuhan has left some wondering why China is not relaxing its virus strategy now that the vast majority of its one billion people have been fully vaccinated. They're building more facilities but there is no indication the authorities plan to ease the restrictions that have effectively ended international travel for people in China.

Feature selection techniques are essential for high-dimensional data analysis. In the last two decades, their popularity has been fuelled by the increasing availability of high-throughput biomolecular data where high-dimensionality is a common data property. Recent advances in biotechnologies enable global profiling of various molecular and cellular features at single-cell resolution, resulting in large-scale datasets with increased complexity. These technological developments have led to a resurgence in feature selection research and application in the single-cell field. Here, we revisit feature selection techniques and summarise recent developments. We review their versatile application to a range of single-cell data types including those generated from traditional cytometry and imaging technologies and the latest array of single-cell omics technologies. We highlight some of the challenges and future directions on which feature selection could have a significant impact. Finally, we consider the scalability and make general recommendations on the utility of each type of feature selection method. We hope this review serves as a reference point to stimulate future research and application of feature selection in the single-cell era.

Network load balancers are important components in data centers to provide scalable services. Workload distribution algorithms are based on heuristics, e.g., Equal-Cost Multi-Path (ECMP), Weighted-Cost Multi-Path (WCMP) or naive machine learning (ML) algorithms, e.g., ridge regression. Advanced ML-based approaches help achieve performance gain in different networking and system problems. However, it is challenging to apply ML algorithms on networking problems in real-life systems. It requires domain knowledge to collect features from low-latency, high-throughput, and scalable networking systems, which are dynamic and heterogenous. This paper proposes Aquarius to bridge the gap between ML and networking systems and demonstrates its usage in the context of network load balancers. This paper demonstrates its ability of conducting both offline data analysis and online model deployment in realistic systems. The results show that the ML model trained and deployed using Aquarius improves load balancing performance yet they also reveals more challenges to be resolved to apply ML for networking systems.

I consider the setting where reviewers offer very noisy scores for a number of items for the selection of high-quality ones (e.g., peer review of large conference proceedings) whereas the owner of these items knows the true underlying scores but prefers not to provide this information. To address this withholding of information, in this paper, I introduce the \textit{Isotonic Mechanism}, a simple and efficient approach to improving on the imprecise raw scores by leveraging certain information that the owner is incentivized to provide. This mechanism takes as input the ranking of the items from best to worst provided by the owner, in addition to the raw scores provided by the reviewers. It reports adjusted scores for the items by solving a convex optimization problem. Under certain conditions, I show that the owner's optimal strategy is to honestly report the true ranking of the items to her best knowledge in order to maximize the expected utility. Moreover, I prove that the adjusted scores provided by this owner-assisted mechanism are indeed significantly more accurate than the raw scores provided by the reviewers. This paper concludes with several extensions of the Isotonic Mechanism and some refinements of the mechanism for practical considerations.

The predominant approach for language modeling is to process sequences from left to right, but this eliminates a source of information: the order by which the sequence was generated. One strategy to recover this information is to decode both the content and ordering of tokens. Existing approaches supervise content and ordering by designing problem-specific loss functions and pre-training with an ordering pre-selected. Other recent works use iterative search to discover problem-specific orderings for training, but suffer from high time complexity and cannot be efficiently parallelized. We address these limitations with an unsupervised parallelizable learner that discovers high-quality generation orders purely from training data -- no domain knowledge required. The learner contains an encoder network and decoder language model that perform variational inference with autoregressive orders (represented as permutation matrices) as latent variables. The corresponding ELBO is not differentiable, so we develop a practical algorithm for end-to-end optimization using policy gradients. We implement the encoder as a Transformer with non-causal attention that outputs permutations in one forward pass. Permutations then serve as target generation orders for training an insertion-based Transformer language model. Empirical results in language modeling tasks demonstrate that our method is context-aware and discovers orderings that are competitive with or even better than fixed orders.