Enabling efficient, high-performance data access in parallel file systems (PFS) is critical for today's highperformance computing systems. PFS client-side I/O heavily impacts the final I/O performance delivered to individual applications and the entire system. Autotuning the key client-side I/O behaviors has been extensively studied and shows promising results. However, existing work has heavily relied on extensive number of global runtime metrics to monitor and accurate modeling of applications' I/O patterns. Such heavy overheads significantly limit the ability to enable fine-grained, dynamic tuning in practical systems. In this study, we propose DIAL (Decentralized I/O AutoTuning via Learned Client-side Local Metrics) which takes a drastically different approach. Instead of trying to extract the global I/O patterns of applications, DIAL takes a decentralized approach, treating each I/O client as an independent unit and tuning configurations using only its locally observable metrics. With the help of machine learning models, DIAL enables multiple tunable units to make independent but collective decisions, reacting to what is happening in the global storage systems in a timely manner and achieving better I/O performance globally for the application.

We propose a pad-focused, net-by-net, two-stage printed circuit board (PCB) routing approach comprising the global routing using Monte Carlo tree search (MCTS) and the detailed routing using$\mathrm{A}^{*}$:. Compared with conventional PCB routing algorithms, our approach can route PCB components in both BGA and non-BGA packages. To minimize the gap between the global and detailed routing stages, a polygon-based dynamic routable region partitioning mechanism is introduced. Experimental results show that our approach outperforms state-of-the-art routers such as DeepPCB and FreeRouting in terms of success rate or wirelength.

Evaluating machine-generated summaries without a human-written reference summary has been a need for a long time. Inspired by preference labeling in existing work of summarization evaluation, we propose to judge summary quality by learning the preference rank of summaries using the Bradley-Terry power ranking model from inferior summaries generated by corrupting base summaries. Extensive experiments on several datasets show that our weakly supervised scheme can produce scores highly correlated with human ratings.

Forrest Bao, Ge Luo, Hebi Li, Minghui Qiu, Yinfei Yang, Youbiao He, Cen Chen. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Graphical PCB design tools like KiCAD lack support for high-level abstraction such as functions and loops. To improve PCB design productivity, we hereby present BHDL, a programming framework for PCB designs. In its compact and declarative syntax, schematics and layouts can be modeled effectively and expressed concisely. Treating all circuits, even a resistor, as functions, BHDL naturally supports modularized development that builds a complex design up from smaller designs hierarchically. As an embedded Domain Specific Language (eDSL), BHDL allows users to leverage the full feature of the host language for customization and extension. Our Jupyter kernel supports web-based, REPL-style development and generates auto-placed PCBs.

Today's high-performance computing (HPC) platforms are still dominated by batch jobs. Accordingly, effective batch job scheduling is crucial to obtain high system efficiency. Existing HPC batch job schedulers typically leverage heuristic priority functions to prioritize and schedule jobs. But, once configured and deployed by the experts, such priority functions can hardly adapt to the changes of job loads, optimization goals, or system settings, potentially leading to degraded system efficiency when changes occur. To address this fundamental issue, we present RLScheduler, an automated HPC batch job scheduler built on reinforcement learning. RLScheduler relies on minimal manual interventions or expert knowledge, but can learn high-quality scheduling policies via its own continuous `trial and error'. We introduce a new kernel-based neural network structure and trajectory filtering mechanism in RLScheduler to improve and stabilize the learning process. Through extensive evaluations, we confirm that RLScheduler can learn high-quality scheduling policies towards various workloads and various optimization goals with relatively low computation cost. Moreover, we show that the learned models perform stably even when applied to unseen workloads, making them practical for production use.

Product search and recommendation is a task that every e-commerce platform wants to outperform their peels on. However, training a good search or recommendation model often requires more data than what many platforms have. Fortunately, the search tasks on different platforms share the common underlying structure. Considering each platform as a domain, we propose a cross-domain learning approach to help the task on data-deficient platforms by leveraging the data from data-abundant platforms. In our solution, the importance of features in different domains is addressed by a domain-specific attention network. Meanwhile, a multi-task regularizer based on Wasserstein distance is introduced to help extract both domain-invariant and domain-specific features. Our model consistently outperforms the competing methods on both public and real-world industry datasets. Quantitative evaluation shows that our model can discover important features for different domains, which helps us better understand different user needs across platforms. Last but not least, we have deployed our model online in three big e-commerce platforms namely Taobao, Tmall, and Qintao, and observed better performance than the production models for all the platforms.

A customer service bot is now a necessary component of an e-commerce platform. As a core module of the customer service bot, user intent prediction can help predict user questions before they ask. A typical solution is to find top candidate questions that a user will be interested in. Such solution ignores the inter-relationship between questions and often aims to maximize the immediate reward such as clicks, which may not be ideal in practice. Hence, we propose to view the problem as a sequential decision making process to better capture the long-term effects of each recommendation in the list. Intuitively, we formulate the problem as a Markov decision process and consider using reinforcement learning for the problem. With this approach, questions presented to users are both relevant and diverse. Experiments on offline real-world dataset and online system demonstrate the effectiveness of our proposed approach.

Consumers today face too many reviews to read when shopping online. Presenting the most helpful reviews, instead of all, to them will greatly ease purchase decision making. Most of the existing studies on review helpfulness prediction focused on domains with rich labels, not suitable for domains with insufficient labels. In response, we explore a multi-domain approach that learns domain relationships to help the task by transferring knowledge from data-rich domains to data-deficient domains. To better model domain differences, our approach gates multi-granularity embeddings in a Neural Network (NN) based transfer learning framework to reflect the domain-variant importance of words. Extensive experiments empirically demonstrate that our model outperforms the state-of-the-art baselines and NN-based methods without gating on this task. Our approach facilitates more effective knowledge transfer between domains, especially when the target domain dataset is small. Meanwhile, the domain relationship and domain-specific embedding gating are insightful and interpretable.

Mindboggle (http://mindboggle.info) is an open source brain morphometry platform that takes in preprocessed T1-weighted MRI data and outputs volume, surface, and tabular data containing label, feature, and shape information for further analysis. In this article, we document the software and demonstrate its use in studies of shape variation in healthy and diseased humans. The number of different shape measures and the size of the populations make this the largest and most detailed shape analysis of human brains ever conducted. Brain image morphometry shows great potential for providing much-needed biological markers for diagnosing, tracking, and predicting progression of mental health disorders. Very few software algorithms provide more than measures of volume and cortical thickness, while more subtle shape measures may provide more sensitive and specific biomarkers. Mindboggle computes a variety of (primarily surface-based) shapes: area, volume, thickness, curvature, depth, Laplace-Beltrami spectra, Zernike moments, etc. We evaluate Mindboggle's algorithms using the largest set of manually labeled, publicly available brain images in the world and compare them against state-of-the-art algorithms where they exist. All data, code, and results of these evaluations are publicly available.

Product reviews greatly influence purchase decisions in online shopping. A common burden of online shopping is that consumers have to search for the right answers through massive reviews, especially on popular products. Hence, estimating and predicting the helpfulness of reviews become important tasks to directly improve shopping experience. In this paper, we propose a new approach to helpfulness prediction by leveraging aspect analysis of reviews. Our hypothesis is that a helpful review will cover many aspects of a product at different emphasis levels. The first step to tackle this problem is to extract proper aspects. Because related products share common aspects to different degrees, we propose an aspect extraction model making use of product category information to balance the aspects of a general category and those of subcategories under it. On top of this model, a two-layer regressor is trained for helpfulness prediction. Experiment results show that we can improve helpfulness prediction by 7% than the baseline on 5 popular product categories from Amazon.com.

13C metabolic flux analysis (13C-MFA) has been widely used to measure in vivo enzyme reaction rates (i.e., metabolic flux) in microorganisms. Mining the relationship between environmental and genetic factors and metabolic fluxes hidden in existing fluxomic data will lead to predictive models that can significantly accelerate flux quantification. In this paper, we present a web-based platform MFlux (http://mflux.org) that predicts the bacterial central metabolism via machine learning, leveraging data from approximately 100 13C-MFA papers on heterotrophic bacterial metabolisms. Three machine learning methods, namely Support Vector Machine (SVM), k-Nearest Neighbors (k-NN), and Decision Tree, were employed to study the sophisticated relationship between influential factors and metabolic fluxes. We performed a grid search of the best parameter set for each algorithm and verified their performance through 10-fold cross validations. SVM yields the highest accuracy among all three algorithms. Further, we employed quadratic programming to adjust flux profiles to satisfy stoichiometric constraints. Multiple case studies have shown that MFlux can reasonably predict fluxomes as a function of bacterial species, substrate types, growth rate, oxygen conditions, and cultivation methods. Due to the interest of studying model organism under particular carbon sources, bias of fluxome in the dataset may limit the applicability of machine learning models. This problem can be resolved after more papers on 13C-MFA are published for non-model species.

Boolean SATisfiability (SAT) is an important problem in AI. SAT solvers have been effectively used in important industrial applications including automated planning and verification. In this paper, we present novel algorithms for fast SAT solving by employing two common subclause elimination (CSE) approaches. Our motivation is that modern SAT solving techniques can be more efficient on CSE-processed instances. Empirical study shows that CSE can significantly speed up SAT solving.

Electroencephalogram (EEG) is a gold standard in epilepsy diagnosis and has been widely studied for epilepsy-related signal classification. In the past few years, discrete wavelet transform (DWT) has been widely used to analyze epileptic EEG. However, there are two practical questions unanswered: 1. what the best mother wavelet for epileptic EEG analysis is; 2. what the optimal level of wavelet decomposition is. The main challenge in using wavelet transform is selecting the optimal mother wavelet for the given task, as different mother wavelet applied on the same signal may produces different results. Such a problem also exist in epileptic EEG analysis based on wavelet. Deeper DWT can yield more detailed depiction of signals but it requires substantially more computational time. In this paper, we study these problems, using the most common epileptic EEG classification task, seizure detection, as an example. The results show that all 7 mother wavelets used in this work achieve high seizure detection accuracy at high decomposition levels. Also, decomposition level effects the detection accuracy more significantly than mother wavelets. For all wavelets, decomposition beyond level 7 improves accuracy limitedly and even decreases accuracy. We further study the most effective bands and features for seizure detection. An interpretation to our results is that seizure and non-seizure EEGs differ across all conventional frequency bands of human EEG rhythms. The best accuracy of seizure detection achieved in this research is 92.30% using coif3 from levels 2 to 7.

Like any communication networks, Wireless ad hoc networks (WANETs) require routing to support many applications such as data aggregation and over-the- air firmware update. Traditional route discovery in WANETs floods request all over the network causing broadcast storm that will heavily consume the precious power resources on nodes. To respond, gossip routing is proposed to reduce the number of messages. In recent years, several algorithms have been developed to improve its efficiency using location information. However, they only make use of distance but not directional information which is much cheaper to acquire in WANETs. In this paper, we develop an approach to improve location- aided routing by making use of directional information, with and without distance information. Empirical results show that our approach outperforms existing gossip routing algorithms, including location-aided ones, on random and deployed WSNs. When only general direction information of nodes is given, performances of our algorithms only drop slightly. Hoping this approach can also benefit other types of routing tasks, we also migrate the spirit of this algorithm onto opportunistic routing and show improved performance than existing location-aided opportunistic routing.

Locating lossy nodes in wireless sensor networks (WSNs) is difficult due to the large amount of sensor nodes, and their limited resources. The state-of-the-art work frames lossy node localization in WSNs as an optimal sequential testing problem guided by end-to-end data. It combines both active and passive measurements to minimize testing cost and number of iterations. However, this hybrid approach has many limitations. Inspired by the success of statistic methods in coverage-based software testing, and the similarity between software testing and lossy node localization, we develop an improved approach by employing Chi-square test in WSN lossy node localization. Supported by well-established statistic theories, our elegant approach delivers great performance. Experiments on randomly generated networks and deployed networks show significant performance improvement using the proposed algorithm. We expect to use this approach for other diagnostic problems in WSNs.

Recently, a new language AC(C) was proposed to integrate answer set programming (ASP) and constraint logic programming (CLP). In this paper, we show that temporally expressive planning problems in PDDL2.1 can be translated into AC(C) and solved using AC(C) solvers. Compared with existing approaches, the new approach puts less restrictions on the planning problems and is easy to extend with new features like PDDL axioms. It can also leverage the inference engine for AC(C) which has the potential to exploit the best reasoning mechanisms developed in the ASP, SAT and CP communities.

Key challenges to widespread deployment of mobile robots to interact with humans in real-world domains include the ability to: (a) robustly represent and revise domain knowledge; (b) autonomously adapt sensing and processing to the task at hand; and (c) learn from unreliable high-level human feedback. Partially observable Markov decision processes (POMDPs) have been used to plan sensing and navigation in different application domains. It is however a challenge to include common sense knowledge obtained from sensory or human inputs in POMDPs. In addition, information extracted from sensory and human inputs may have varying levels of relevance to current and future tasks. On the other hand, although a non-monotonic logic programming paradigm such as Answer Set Programming (ASP) is wellsuited for common sense reasoning, it is unable to model the uncertainty in real-world sensing and navigation (Gelfond 2008). This paper presents a hybrid framework that integrates ASP, hierarchical POMDPs (Zhang and Sridharan 2012) and psychophysics principles to address the challenges stated above. Experimental results in simulation and on mobile robots deployed in indoor domains show that the framework results in reliable and efficient operation.

A collection of sets may have some interesting properties which help identify efficient algorithms for constraint satisfaction problems and combinatorial auction problems. One of the properties is tree convexity. A collection S of sets is tree convex if we can find a tree T whose nodes are the union of the sets of S and each set of S is the nodes of a subtree of T. This concept extends that of row convex sets each of which is an interval over a total ordering of the elements of the union of these sets. An interesting problem is to find efficient algorithms to test whether a collection of sets is tree convex. It is not known before if there exists a linear time algorithm for this test. In this paper, we review the materials that are the key to a linear algorithm: hypergraphs, a characterization of tree convex sets and the acyclic hypergraph test algorithm. Some typos in the original paper of the acyclicity test are corrected here. Some experiments show that the linear algorithm is significantly faster than a well‐known existing algorithm.

Medical treatment decision making is a good application of knowledge representation and reasoning. We are particularly interested in using it to resolve treatment conflicts, a complicated condition when two treatments cannot be given simultaneously to a patient of multiple symptoms. The logic system is required to reason on cases with and without treatment conflicts. Thanks to the nonmonotonicity of Answer Set Programming (ASP), we elegantly automate medical treatment conflict resolving on an example problem and show the importance of nonmonotonicity in medical reasoning.

Counting phased small RNA cycles (PSRC) from mapped small RNA positions is a repeatedly invoked subproblem in the computation of identifying TRANS-ACTING siRNA (TAS) loci and loci of other small RNAs forming through mechanisms similar to that of trans-acting small interfering RNAs (ta-siRNAs). The efficiency of counting PSRC has a clear impact on the efficiency of the algorithms predicting these loci. There are two closely related variants on counting PSRC in real applications: WPSRC, which counts the number of distinct small RNAs falling onto the phased positions in a sliding window, and MPSRC, which counts the maximum consecutive PSRC from mapped small RNA positions. In this paper, we develop fast algorithms for both WPSRC and MPSRC. Our algorithms have O(max(S)) time complexity, while the existing algorithm and its variant have O(|S|·max(S)) and O(|S|·L) time complexity for MPSRC and WPSRC respectively, where S is a set of mapped small RNA positions and L the length of sliding window for WPSRC. Experimental results on two real-life datasets show that our algorithms are significantly faster than the existing algorithm and its variant. The proposed algorithms are applicable to TAS-like clusters with any PSRC length including 21-nt.

Epilepsy is one of the most common neurological disorders that greatly impair patients' daily lives. Traditional epileptic diagnosis relies on tedious visual screening by neurologists from lengthy EEG recording that requires the presence of seizure (ictal) activities. Nowadays, there are many systems helping the neurologists to quickly find interesting segments from the lengthy signal by automatic seizure detection. However, we notice that it is very difficult, if not impossible, to obtain long-term EEG data with seizure activities for epilepsy patients in areas lack of medical resources and trained neurologists. Therefore, we propose to study automated epileptic diagnosis using interictal EEG data that is much easier to collect than ictal data. The authors are not aware of any report on automated EEG diagnostic system that can accurately distinguish patients' interictal EEG from the EEG of normal people. The research presented in this paper, therefore, aims to develop an automated diagnostic system that can use interictal EEG data to diagnose whether the person is epileptic. Such a system should also detect seizure activities for further investigation by doctors and potential patient monitoring. To develop such a system, we extract three classes of features from the EEG data and build a probabilistic neural network (PNN) fed with these features. Leave-one-out cross-validation (LOO-CV) on a widely used epileptic-normal data set reflects an impressive 99.3% accuracy of our system on distinguishing normal people's EEG from patients' interictal EEG. We also find our system can be used in patient monitoring (seizure detection) and seizure focus localization, with 96.7% and 76.5% accuracy respectively on the data set.

Facial expressions are considered a critical factor in neonatal pain assessment. This paper proposes a pain expression recognition method using boosted Gabor features. Each neonatal facial image is convoluted with the 2D Gabor filters to extract 412,160 Gabor features. Since the high-dimension Gabor feature vectors are quite redundant, we employs a modified version of AdaBoost algorithm to select and combine the most informative features for classification. The "pain vs. non-pain" problem is treated as two sub-problems by using a coarse-to-fine hierarchical classifier. Experiments with 510 neonatal expression images show that the proposed method is quite effective. Only 30 Gabor features are enough to achieve good classification performance. The recognition rate of pain versus non-pain is up to 88% (i.e. error rate isin=0.12). Compared with one existing algorithm for neonatal facial pain recognition,our approach can reach similar accuracy in much lower time complexity.

In the process of training support vector machines (SVMs) by decomposition methods, working set selection is an important technique, and some exciting schemes were employed into this field. To improve working set selection, we propose a new model for working set selection in sequential minimal optimization (SMO) decomposition methods. In this model, it selects B as working set without reselection. Some properties are given by simple proof, and experiments demonstrate that the proposed method is in general faster than existing methods.