One-shot Manner Research Articles

Privacy-preserving large language models for structured medical information retrieval

Most clinical information is encoded as free text, not accessible for quantitative analysis. This study presents an open-source pipeline using the local large language model (LLM) “Llama 2” to extract quantitative information from clinical text and evaluates its performance in identifying features of decompensated liver cirrhosis. The LLM identified five key clinical features in a zero- and one-shot manner from 500 patient medical histories in the MIMIC IV dataset. We compared LLMs of three sizes and various prompt engineering approaches, with predictions compared against ground truth from three blinded medical experts. Our pipeline achieved high accuracy, detecting liver cirrhosis with 100% sensitivity and 96% specificity. High sensitivities and specificities were also yielded for detecting ascites (95%, 95%), confusion (76%, 94%), abdominal pain (84%, 97%), and shortness of breath (87%, 97%) using the 70 billion parameter model, which outperformed smaller versions. Our study successfully demonstrates the capability of locally deployed LLMs to extract clinical information from free text with low hardware requirements.

Non-autoregressive personalized bundle generation

The personalized bundle generation problem, which aims to create a preferred bundle for user from numerous candidate items, receives increasing attention in recommendation. However, existing works ignore the order-invariant nature of the bundle and adopt sequential modeling methods as the solution, which might introduce inductive bias and cause a large latency in prediction. To address this problem, we propose to perform the bundle generation via non-autoregressive mechanism and design a novel encoder–decoder framework named BundleNAT, which can effectively output the targeted bundle in one-shot without relying on any inherent order. In detail, instead of learning sequential dependency, we propose to adopt pre-training techniques and graph neural network to fully embed user-based preference and item-based compatibility information, and use a self-attention based encoder to further extract global dependency pattern. We then design a permutation-equivariant decoding architecture that is able to directly output the desired bundle in a one-shot manner. Experiments on three real-world datasets from Youshu and Netease show the proposed BundleNAT significantly outperforms the current state-of-the-art methods in average by up to 35.92%, 10.97% and 23.67% absolute improvements in Precision, Precision+, and Recall, respectively.

Cross-Gate MLP with Protein Complex Invariant Embedding Is a One-Shot Antibody Designer

Antibodies are crucial proteins produced by the immune system in response to foreign substances or antigens. The specificity of an antibody is determined by its complementarity-determining regions (CDRs), which are located in the variable domains of the antibody chains and form the antigen-binding site. Previous studies have utilized complex techniques to generate CDRs, but they suffer from inadequate geometric modeling. Moreover, the common iterative refinement strategies lead to an inefficient inference. In this paper, we propose a simple yet effective model that can co-design 1D sequences and 3D structures of CDRs in a one-shot manner. To achieve this, we decouple the antibody CDR design problem into two stages: (i) geometric modeling of protein complex structures and (ii) sequence-structure co-learning. We develop a novel macromolecular structure invariant embedding, typically for protein complexes, that captures both intra- and inter-component interactions among the backbone atoms, including Calpha, N, C, and O atoms, to achieve comprehensive geometric modeling. Then, we introduce a simple cross-gate MLP for sequence-structure co-learning, allowing sequence and structure representations to implicitly refine each other. This enables our model to design desired sequences and structures in a one-shot manner. Extensive experiments are conducted to evaluate our results at both the sequence and structure level, which demonstrate that our model achieves superior performance compared to the state-of-the-art antibody CDR design methods.

OffsetNet: Towards Efficient Multiple Object Tracking, Detection, and Segmentation.

Offset-based representation has emerged as a promising approach for modeling semantic relations between pixels and object motion, demonstrating efficacy across various computer vision tasks. In this paper, we introduce a novel one-stage multi-tasking network tailored to extend the offset-based approach to MOTS. Our proposed framework, named OffsetNet, is designed to concurrently address amodal bounding box detection, instance segmentation, and tracking. It achieves this by formulating these three tasks within a unified pixel-offset-based representation, thereby achieving excellent efficiency and encouraging mutual collaborations. OffsetNet achieves several remarkable properties: first, the encoder is empowered by a novel Memory Enhanced Linear Self-Attention (MELSA) block to efficiently aggregate spatial-temporal features; second, all tasks are decoupled fairly using three lightweight decoders that operate in a one-shot manner; third, a novel cross-frame offsets prediction module is proposed to enhance the robustness of tracking against occlusions. With these merits, OffsetNet achieves 76.83% HOTA on KITTI MOTS benchmark, which is the best result without relying on 3D detection. Furthermore, OffsetNet achieves 74.83% HOTA at 50 FPS on the KITTI MOT benchmark, which is nearly 3.3 times faster than CenterTrack with better performance. We hope our approach will serve as a solid baseline and encourage future research in this field.

Extending the Omniglot Challenge: Imitating Handwriting Styles on a New Sequential Data Set

With the increasing effectiveness of one-/few-shot learning techniques in the context of handwritten character generation and recognition, the call to extend the commonly associated Omniglot challenge is becoming more pressing. However, the sequential Omniglot dataset represents unrealistically written characters. Therefore, we present new data, a new challenge, and a new model as follows: On the data side, we introduce DigiLeTs, a dataset containing 23,870 new trajectories of Latin letters and Arabic numbers from 77 participants with high natural variance within character types. On the challenge side, we introduce the task to imitate handwriting styles in a one-shot manner. On the model side, we extend a generative, recurrent neural network model equipped with a one-shot inference mechanism that allows to reuse previously extracted compositional encodings and that has already been proven promising to solve the original Omniglot challenge. The new model is able to reassemble previously learned components into new characters of new styles in a one-shot manner. Most surprisingly, in a strictly forward manner, it often generates plausible, unknown characters in an already known style. With this work, we hope to inspire future research to investigate how compositional structures develop and are employed for rapid, concept-oriented learning, imitation, and understanding.

CCGnet: A deep learning approach to predict Nash equilibrium of chance-constrained games

This paper proposes a novel method for efficiently finding the Nash equilibrium in a chance-constrained game (CCG). Conventional numerical solution methods require significant computational time when solving multiple instances of CCG. We introduce CCGnet, a deep learning approach that is capable of efficiently solving multiple instances of CCG in a one-shot manner. CCGnet employs a specialized network structure and training algorithm based on neurodynamic optimization. We demonstrate the strong performance of CCGnet in practice and show that our proposed method outperforms conventional methods.

A deep learning approach for solving linear programming problems

Finding the optimal solution to a linear programming (LP) problem is a long-standing computational problem in Operations Research. This paper proposes a deep learning approach in the form of feed-forward neural networks to solve the LP problem. The latter is first modeled by an ordinary differential equations (ODE) system, the state solution of which globally converges to the optimal solution of the LP problem. A neural network model is constructed as an approximate state solution to the ODE system, such that the neural network model contains the prediction of the LP problem. Furthermore, we extend the capability of the neural network by taking the parameter of LP problems as an input variable so that one neural network can solve multiple LP instances in a one-shot manner. Finally, we validate the proposed method through a collection of specific LP examples and show concretely how the proposed method solves the example.

Recurrent Tissue-Aware Network for Deformable Registration of Infant Brain MR Images.

Deformable registration is fundamental to longitudinal and population-based image analyses. However, it is challenging to precisely align longitudinal infant brain MR images of the same subject, as well as cross-sectional infant brain MR images of different subjects, due to fast brain development during infancy. In this paper, we propose a recurrently usable deep neural network for the registration of infant brain MR images. There are three main highlights of our proposed method. (i) We use brain tissue segmentation maps for registration, instead of intensity images, to tackle the issue of rapid contrast changes of brain tissues during the first year of life. (ii) A single registration network is trained in a one-shot manner, and then recurrently applied in inference for multiple times, such that the complex deformation field can be recovered incrementally. (iii) We also propose both the adaptive smoothing layer and the tissue-aware anti-folding constraint into the registration network to ensure the physiological plausibility of estimated deformations without degrading the registration accuracy. Experimental results, in comparison to the state-of-the-art registration methods, indicate that our proposed method achieves the highest registration accuracy while still preserving the smoothness of the deformation field. The implementation of our proposed registration network is available online https://github.com/Barnonewdm/ACTA-Reg-Net.

Short-Term Load Forecasting Using Encoder-Decoder WaveNet: Application to the French Grid

The prediction of time series data applied to the energy sector (prediction of renewable energy production, forecasting prosumers’ consumption/generation, forecast of country-level consumption, etc.) has numerous useful applications. Nevertheless, the complexity and non-linear behaviour associated with such kind of energy systems hinder the development of accurate algorithms. In such a context, this paper investigates the use of a state-of-art deep learning architecture in order to perform precise load demand forecasting 24-h-ahead in the whole country of France using RTE data. To this end, the authors propose an encoder-decoder architecture inspired by WaveNet, a deep generative model initially designed by Google DeepMind for raw audio waveforms. WaveNet uses dilated causal convolutions and skip-connection to utilise long-term information. This kind of novel ML architecture presents different advantages regarding other statistical algorithms. On the one hand, the proposed deep learning model’s training process can be parallelized in GPUs, which is an advantage in terms of training times compared to recurrent networks. On the other hand, the model prevents degradations problems (explosions and vanishing gradients) due to the residual connections. In addition, this model can learn from an input sequence to produce a forecast sequence in a one-shot manner. For comparison purposes, a comparative analysis between the most performing state-of-art deep learning models and traditional statistical approaches is presented: Autoregressive-Integrated Moving Average (ARIMA), Long-Short-Term-Memory, Gated-Recurrent-Unit (GRU), Multi-Layer Perceptron (MLP), causal 1D-Convolutional Neural Networks (1D-CNN) and ConvLSTM (Encoder-Decoder). The values of the evaluation indicators reveal that WaveNet exhibits superior performance in both forecasting accuracy and robustness.

<i>run</i>Data: Re-distributing Data via Piggybacking for Geo-distributed Data Analytics over Edges

Efficiently analyzing geo-distributed datasets is emerging as a major demand in a cloud-edge system. Since the datasets are often generated in closer proximity to end users, traditional works mainly focus on offloading proper tasks from those hotspot edges to the datacenter to decrease the overall completion time of submitted jobs in a one-shot manner. However, optimizing the completion time of current job alone is insufficient in a long-term scope since some datasets would be used multiple times. Instead, optimizing the data distribution is much more efficient and could directly benefit forthcoming jobs, although it may postpone the execution of current one. Unfortunately, due to the throwaway feature of data fetcher, existing data analytics systems fail to re-distribute corresponding data out of hotspot edges after the execution of data analytics. In order to minimize the overall completion time for a sequence of jobs as well as to guarantee the performance of current one, we propose to re-distribute the data along with task offloading, and formulate corresponding ε-bounded data-driven task scheduling problem over wide area network under the consideration of edge heterogeneity. We design an online schema <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">run</i> Data, which offloads proper tasks and related data via piggybacking to the datacenter based on delicately calculated probabilities. Through rigorous theoretical analysis, <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">run</i> Data is proved concentrated on its optimum with high probability. We implement <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">run</i> Data based on Spark and HDFS. Both testbed results and trace-driven simulations show that run Data re-distributes proper data via piggybacking and achieves up to 37 percent reduction on average response time compared with state-of-the-art schemas.

Bayesian Automatic Model Compression

Model compression has drawn great attention in deep learning community. A core problem in model compression is to determine the layer-wise optimal compression policy, e.g., the layer-wise bit-width in network quantization. Conventional hand-crafted heuristics rely on human experts and are usually sub-optimal, while recent reinforcement learning based approaches can be inefficient during the exploration of the policy space. In this article, we propose Bayesian automatic model compression (BAMC), which leverages non-parametric Bayesian methods to learn the optimal quantization bit-width for each layer of the network. BAMC is trained in a one-shot manner, avoiding the back and forth (re)-training in reinforcement learning based approaches. Experimental results on various datasets validate that our proposed methods can find reasonable quantization policies efficiently with little accuracy drop for the quantized network.

Neural Source-Filter Waveform Models for Statistical Parametric Speech Synthesis

Neural waveform models have demonstrated better performance than conventional vocoders for statistical parametric speech synthesis. One of the best models, called WaveNet, uses an autoregressive (AR) approach to model the distribution of waveform sampling points, but it has to generate a waveform in a time-consuming sequential manner. Some new models that use inverse-autoregressive flow (IAF) can generate a whole waveform in a one-shot manner but require either a larger amount of training time or a complicated model architecture plus a blend of training criteria. As an alternative to AR and IAF-based frameworks, we propose a neural source-filter (NSF) waveform modeling framework that is straightforward to train and fast to generate waveforms. This framework requires three components to generate waveforms: a source module that generates a sine-based signal as excitation, a non-AR dilated-convolution-based filter module that transforms the excitation into a waveform, and a conditional module that pre-processes the input acoustic features for the source and filter modules. This framework minimizes spectral-amplitude distances for model training, which can be efficiently implemented using short-time Fourier transform routines. As an initial NSF study, we designed three NSF models under the proposed framework and compared them with WaveNet using our deep learning toolkit. It was demonstrated that the NSF models generated waveforms at least 100 times faster than our WaveNet-vocoder, and the quality of the synthetic speech from the best NSF model was comparable to that from WaveNet on a large single-speaker Japanese speech corpus.

Quasi-boundary value methods for regularizing the backward parabolic equation under the optimal control framework**Jun Liu’s research was supported by a Seed Grants for the Transitional and Exploratory Projects (STEP) Award (FY2019) from the SIUE Graduate School. Mingqing Xiao’s research was supported in part by the NSF-DMS140928, 1854638 of the United States.

In this paper, we propose to solve the classical backward parabolic equation under the optimal control framework associated with the Tikhonov regularization formulation, where the initial condition is set to be the control variable while the misfit of the final condition is formulated as the Tikhonov objective functional to be minimized. The corresponding first-order necessary optimality system is discretized in a one-shot manner by a second-order finite difference scheme in space and time. The proposed optimal control setting, as discussed in the paper, provides more general framework in terms of solving this type of ill-posed inverse problem. In particular, several existing nonlocal quasi-boundary value methods appear to be the special cases of our proposed optimal control framework with appropriately chosen generalized regularization setting respectively. The relationship between our Tikhonov regularization approach and the quasi-boundary value methods is discussed in detail. The optimal control approach based on Tikhonov regularization is shown to deliver the known optimal order convergence rate. Numerical examples are provided to demonstrate both effective accuracy of approximation as well as excellent stability under the optimal control realization, comparing with three recent quasi-boundary value methods in literature.

Sequential Saliency Guided Deep Neural Network for Joint Mitosis Identification and Localization in Time-Lapse Phase Contrast Microscopy Images.

The analysis of cell mitotic behavior plays important role in many biomedical research and medical diagnostic applications. To improve the accuracy of mitosis detection in automated analysis systems, this paper proposes the sequential saliency guided deep neural network (SSG-DNN) to jointly identify and localize mitotic events in time-lapse phase contrast microscopy images. It consists of three key modules. First, the module of visual context learning extracts static visual feature and dynamic visual transition within individual volumetric cell regions. Secondly, with these information, the module of sequential saliency modeling aims to discover the saliency distribution over all successive frames in each volumetric region. Finally, the module of sequence structure modeling can leverage both visual context and saliency distribution for mitosis identification and localization. SSG-DNN can jointly realize visual feature learning and sequential structure modeling in the end-to-end framework. Moreover, the proposed method is independent of complicated preconditioning methods for mitotic candidate extraction and can be applied for mitosis detection in one-shot manner. To our knowledge, it is the first weakly supervised work to realize joint mitosis identification and localization only with sequence-wise labels. In our experiments, we evaluate its performances of both tasks on the popular C3H10 dataset and a novel and large-scale dataset, C2C12-16, which contains much more mitotic events and is more challenging owing to diverse cell culture conditions. Experimental results can demonstrate the superiority of the proposed method.

A Machine Learning Approach to Recommending Files in a Collaborative Work Environment

Recommendation of items to users is a problem faced by many companies in a wide spectrum of industries. This problem was traditionally approached in a one-shot manner, such as recommending movies to users based on all the movie ratings observed so far. The evolution of user activity over time was relatively unexplored. This paper presents a Machine Learning approach developed at Box Inc. for making repeated recommendations of files to users in a collaborative work environment. Our results on historical data show that this approach noticeably outperforms the approach currently implemented at Box and also the traditional Matrix Factorization approach.

Multi-UAV Binocular Intersection With One-Shot Communication: Modeling and Algorithms

Binocular intersection is practically indispensable for stereo visual systems that extract 3D information from images of two distinct cameras. This paper proposes a multi-UAV binocular intersection approach where the two distinct cameras are fixed separately on multiple flying unmanned aerial vehicles (UAVs), rather than physically connected with each other as in the pre-existing works. Eventually, both modeling and algorithms are developed for the multi-UAV binocular intersection with extremely limited communication. As the formulated model states, once one flying vehicle detects the target and shares related information only in a one-shot manner, the other vehicles steer their pan-tilt units (PTUs) to reconstruct a spatial binocular intersection automatically. The binocular intersection algorithm is further improved from two aspects to maximize the probability of target capture. On one hand, the search scope is narrowed by developing an iterative algorithm with priori terrain height and variance. On the other hand, an approaching navigation law is designed to reduce view occlusion caused by terrain. Extensive simulations with different target speeds and flight altitudes are carried out to evaluate the effectiveness of the proposed method. The results show that the probability of binocular intersection is over 90% and is subject to little fluctuation against episodic parameter changes. Furthermore, the developed algorithms are verified localization-free and time-saving as well, and hereafter are potentially practicable for micro aerial swarm applications with extremely limited onboard computation and communication.

EDF-VD Scheduling of Flexible Mixed-Criticality System With Multiple-Shot Transitions

The existing mixed-criticality (MC) real-time task models assume that once any high-criticality task overruns, all high-criticality jobs execute up to their most pessimistic WCET estimations simultaneously in a one-shot manner. This is very pessimistic in the sense of unnecessary resource overbooking. In this paper, we propose a more generalized mixed-critical real-time task model, called flexible MC model with multiple-shot transitions (FMC-MST), to address this problem. In FMC-MST, high-criticality tasks can transit multiple intermediate levels to handle less pessimistic overruns independently and to nonuniformly scale the deadline on each level. We develop a run-time schedulability analysis for FMC-MST under EDF-VD scheduling, in which a better tradeoff between the penalties of low-criticality tasks and the overruns of high-criticality tasks is achieved to improve the service quality of low-criticality tasks. We also develop a resource optimization technique to find resource-efficient level-insertion configurations for FMC-MST task systems under MC timing constraints. Experiments demonstrate the effectiveness of FMC-MST compared with the state-of-the-art techniques.

MCL: A Cost-Efficient Nonblocking Multicast Interconnection Network

Interconnection networks lie in the heart of all types of parallel architectures, because how processors or memory modules are connected to each other has a significant impact on the scalability, reliability, cost and performance. For example, a nonblocking interconnection network delivers guaranteed path availability to any connection requests, without interference to existing connections. Also, an interconnection network with multicast capability can distribute data from a single source to all the destinations in a one-shot manner, eliminating unnecessary duplications and minimizing communication delays. However, implementing nonblocking multicast networks imposes great challenges to designers because both of them demand high hardware cost. To deal with this problem, in this paper we propose a novel interconnection network, named Multicast Capable Low-cost Network (MCL) , which is both nonblocking for multicast traffic and cost-efficient. We first design the topologies and routing algorithms for MCL, and then prove its nonblocking multicast properties. Most importantly, we show that MCL achieves the lowest hardware cost in terms of asymptotic number of crosspoints. Specifically, the theoretical upper bound on the crosspoints of an $N \times N$ MCL is $O(N \log ^{3.39} N)$ . The explicitly constructed instance of an $N \times N$ MCL has a cost of $O(N^\frac{5}{4})$ and constant delay from the input ports to the output ports. These theoretical and practical costs are both the lowest compared to previous designs which deliver the same performance.

Unified quantum no-go theorems and transforming of quantum pure states in a restricted set

The linear superposition principle in quantum mechanics is essential for several no-go theorems such as the no-cloning theorem, the no-deleting theorem and the no-superposing theorem. It remains an open problem of finding general forbidden principles to unify these results. In this paper, we investigate general quantum transformations forbidden or permitted by the superposition principle for various goals. First, we prove a no-encoding theorem that forbids linearly superposing of an unknown pure state and a fixed state in Hilbert space of finite dimension. Two general forms include the no-cloning theorem, the no-deleting theorem, and the no-superposing theorem as special cases. Second, we provide a unified scheme for presenting perfect and imperfect quantum tasks (cloning and deleting) in a one-shot manner. This scheme may yield to fruitful results that are completely characterized with the linear independence of the input pure states. The generalized upper bounds for the success probability will be proved. Third, we generalize a recent superposing of unknown states with fixed overlaps when multiple copies of the input states are available.

Developing architecture for upgrading I&C systems of an operating nuclear power plant using a quality attribute-driven design method

Abstract This paper presents the architecture for upgrading the instrumentation and control (I&C) systems of a Korean standard nuclear power plant (KSNP) as an operating nuclear power plant. This paper uses the analysis results of KSNP's I&C systems performed in a previous study. This paper proposes a Preparation–Decision–Design–Assessment (PDDA) process that focuses on quality oriented development, as a cyclical process to develop the architecture. The PDDA was motivated from the practice of architecture-based development used in software engineering fields. In the preparation step of the PDDA, the architecture of digital-based I&C systems was setup for an architectural goal. Single failure criterion and determinism were setup for architectural drivers. In the decision step, defense-in-depth, diversity, redundancy, and independence were determined as architectural tactics to satisfy the single failure criterion, and sequential execution was determined as a tactic to satisfy the determinism. After determining the tactics, the primitive digital-based I&C architecture was determined. In the design step, 17 systems were selected from the KSNP's I&C systems for the upgrade and functionally grouped based on the primitive architecture. The overall architecture was developed to show the deployment of the systems. The detailed architecture of the safety systems was developed by applying a 2-out-of-3 voting logic, and the detailed architecture of the non-safety systems was developed by hot-standby redundancy. While developing the detailed architecture, three ways of signal transmission were determined with proper rationales: hardwire, datalink, and network. In the assessment step, the required network performance, considering the worst-case of data transmission was calculated: the datalink was required by 120 kbps, the safety network by 5 Mbps, and the non-safety network by 60 Mbps. The architecture covered 17 systems out of 22 KSNP's I&C systems. The architecture is implementable with the equipment developed in South Korea. The architecture can be used as a model to upgrade the existing I&C systems in a planned, large-scale, and one-shot manner. A more detailed architecture down to software level will be developed in the future.