Abstractive speech summarization (SSUM) aims to generate human-like summaries from speech. Given variations in information captured and phrasing, recordings can be summarized in multiple ways. Therefore, it is more reasonable to consider a probabilistic distribution of all potential summaries rather than a single summary. However, conventional SSUM models are mostly trained and evaluated with a single ground-truth (GT) human-annotated deterministic summary for every recording. Generating multiple human references would be ideal to better represent the distribution statistically, but is impractical because annotation is expensive. We tackle this challenge by proposing AugSumm, a method to leverage large language models (LLMs) as a proxy for human annotators to generate augmented summaries for training and evaluation. First, we explore prompting strategies to generate synthetic summaries from ChatGPT. We validate the quality of synthetic summaries using multiple metrics including human evaluation, where we find that summaries generated using AugSumm are perceived as more valid to humans. Second, we develop methods to utilize synthetic summaries in training and evaluation. Experiments on How2 demonstrate that pre-training on synthetic summaries and fine-tuning on GT summaries improves ROUGE-L by 1 point on both GT and AugSumm-based test sets. AugSumm summaries are available at https://github.com/Jungjee/AugSumm.
In recent years, deep learning has emerged as a potent tool across a multitude of domains, leading to a surge in research pertaining to its application in the wearable human activity recognition (WHAR) domain. Despite the rapid development, concerns have been raised about the lack of standardization and consistency in the procedures used for experimental model training, which may affect the reproducibility and reliability of research results. In this paper, we provide an exhaustive review of contemporary deep learning research in the field of WHAR and collate information pertaining to the training procedure employed in various studies. Our findings suggest that a major trend is the lack of detail provided by model training protocols. Besides, to gain a clearer understanding of the impact of missing descriptions, we utilize a control variables approach to assess the impact of key tunable components (e.g., optimization techniques and early stopping criteria) on the inter-subject generalization capabilities of HAR models. With insights from the analyses, we define a novel integrated training procedure tailored to the WHAR model. Empirical results derived using five well-known \ac{whar} benchmark datasets and three classical HAR model architectures demonstrate the effectiveness of our proposed methodology: in particular, there is a significant improvement in macro F1 leave one subject out cross-validation performance.
Time series forecasting is crucial and challenging in the real world. The recent surge in interest regarding time series foundation models, which cater to a diverse array of downstream tasks, is noteworthy. However, existing methods often overlook the multi-scale nature of time series, an aspect crucial for precise forecasting. To bridge this gap, we propose HiMTM, a hierarchical multi-scale masked time series modeling method designed for long-term forecasting. Specifically, it comprises four integral components: (1) hierarchical multi-scale transformer (HMT) to capture temporal information at different scales; (2) decoupled encoder-decoder (DED) forces the encoder to focus on feature extraction, while the decoder to focus on pretext tasks; (3) multi-scale masked reconstruction (MMR) provides multi-stage supervision signals for pre-training; (4) cross-scale attention fine-tuning (CSA-FT) to capture dependencies between different scales for forecasting. Collectively, these components enhance multi-scale feature extraction capabilities in masked time series modeling and contribute to improved prediction accuracy. We conduct extensive experiments on 7 mainstream datasets to prove that HiMTM has obvious advantages over contemporary self-supervised and end-to-end learning methods. The effectiveness of HiMTM is further showcased by its application in the industry of natural gas demand forecasting.
Eye movement (EM) is a new highly secure biometric behavioral modality that has received increasing attention in recent years. Although deep neural networks, such as convolutional neural network (CNN), have recently achieved promising performance, current solutions fail to capture local and global temporal dependencies within eye movement data. To overcome this problem, we propose in this paper a mixed transformer termed EmMixformer to extract time and frequency domain information for eye movement recognition. To this end, we propose a mixed block consisting of three modules, transformer, attention Long short-term memory (attention LSTM), and Fourier transformer. We are the first to attempt leveraging transformer to learn long temporal dependencies within eye movement. Second, we incorporate the attention mechanism into LSTM to propose attention LSTM with the aim to learn short temporal dependencies. Third, we perform self attention in the frequency domain to learn global features. As the three modules provide complementary feature representations in terms of local and global dependencies, the proposed EmMixformer is capable of improving recognition accuracy. The experimental results on our eye movement dataset and two public eye movement datasets show that the proposed EmMixformer outperforms the state of the art by achieving the lowest verification error.
We propose a methodology, based on machine learning and optimization, for selecting a solver configuration for a given instance. First, we employ a set of solved instances and configurations in order to learn a performance function of the solver. Secondly, we formulate a mixed-integer nonlinear program where the objective/constraints explicitly encode the learnt information, and which we solve, upon the arrival of an unknown instance, to find the best solver configuration for that instance, based on the performance function. The main novelty of our approach lies in the fact that the configuration set search problem is formulated as a mathematical program, which allows us to a) enforce hard dependence and compatibility constraints on the configurations, and b) solve it efficiently with off-the-shelf optimization tools.
The visual feature pyramid has proven its effectiveness and efficiency in target detection tasks. Yet, current methodologies tend to overly emphasize inter-layer feature interaction, neglecting the crucial aspect of intra-layer feature adjustment. Experience underscores the significant advantages of intra-layer feature interaction in enhancing target detection tasks. While some approaches endeavor to learn condensed intra-layer feature representations using attention mechanisms or visual transformers, they overlook the incorporation of global information interaction. This oversight results in increased false detections and missed targets.To address this critical issue, this paper introduces the Global Feature Pyramid Network (GFPNet), an augmented version of PAFPN that integrates global information for enhanced target detection. Specifically, we leverage a lightweight MLP to capture global feature information, utilize the VNC encoder to process these features, and employ a parallel learnable mechanism to extract intra-layer features from the input image. Building on this foundation, we retain the PAFPN method to facilitate inter-layer feature interaction, extracting rich feature details across various levels.Compared to conventional feature pyramids, GFPN not only effectively focuses on inter-layer feature information but also captures global feature details, fostering intra-layer feature interaction and generating a more comprehensive and impactful feature representation. GFPN consistently demonstrates performance improvements over object detection baselines.
Spatio-temporal video grounding (or STVG) task aims at locating a spatio-temporal tube for a specific instance given a text query. Despite advancements, current methods easily suffer the distractors or heavy object appearance variations in videos due to insufficient object information from the text, leading to degradation. Addressing this, we propose a novel framework, context-guided STVG (CG-STVG), which mines discriminative instance context for object in videos and applies it as a supplementary guidance for target localization. The key of CG-STVG lies in two specially designed modules, including instance context generation (ICG), which focuses on discovering visual context information (in both appearance and motion) of the instance, and instance context refinement (ICR), which aims to improve the instance context from ICG by eliminating irrelevant or even harmful information from the context. During grounding, ICG, together with ICR, are deployed at each decoding stage of a Transformer architecture for instance context learning. Particularly, instance context learned from one decoding stage is fed to the next stage, and leveraged as a guidance containing rich and discriminative object feature to enhance the target-awareness in decoding feature, which conversely benefits generating better new instance context for improving localization finally. Compared to existing methods, CG-STVG enjoys object information in text query and guidance from mined instance visual context for more accurate target localization. In our experiments on three benchmarks, including HCSTVG-v1/-v2 and VidSTG, CG-STVG sets new state-of-the-arts in m_tIoU and m_vIoU on all of them, showing its efficacy. The code will be released at https://github.com/HengLan/CGSTVG.
Addressing the complexity of comprehensive information retrieval, this study introduces an innovative, iterative retrieval-augmented generation system. Our approach uniquely integrates a vector-space driven re-ranking mechanism with concurrent brainstorming to expedite the retrieval of highly relevant documents, thereby streamlining the generation of potential queries. This sets the stage for our novel hybrid process, which synergistically combines hypothesis formulation with satisfying decision-making strategy to determine content adequacy, leveraging a chain of thought-based prompting technique. This unified hypothesize-satisfied phase intelligently distills information to ascertain whether user queries have been satisfactorily addressed. Upon reaching this criterion, the system refines its output into a concise representation, maximizing conceptual density with minimal verbosity. The iterative nature of the workflow enhances process efficiency and accuracy. Crucially, the concurrency within the brainstorming phase significantly accelerates recursive operations, facilitating rapid convergence to solution satisfaction. Compared to conventional methods, our system demonstrates a marked improvement in computational time and cost-effectiveness. This research advances the state-of-the-art in intelligent retrieval systems, setting a new benchmark for resource-efficient information extraction and abstraction in knowledge-intensive applications.
Benefitting from the vast spatial degrees of freedom, the amalgamation of integrated sensing and communication (ISAC) and massive multiple-input multiple-output (MIMO) is expected to simultaneously improve spectral and energy efficiencies as well as the sensing capability. However, a large number of antennas deployed in massive MIMO-ISAC raises critical challenges in acquiring both accurate channel state information and target parameter information. To overcome these two challenges with a unified framework, we first analyze their underlying system models and then propose a novel tensor-based approach that addresses both the channel estimation and target sensing problems. Specifically, by parameterizing the high-dimensional communication channel exploiting a small number of physical parameters, we associate the channel state information with the sensing parameters of targets in terms of angular, delay, and Doppler dimensions. Then, we propose a shared training pattern adopting the same time-frequency resources such that both the channel estimation and target parameter estimation can be formulated as a canonical polyadic decomposition problem with a similar mathematical expression. On this basis, we first investigate the uniqueness condition of the tensor factorization and the maximum number of resolvable targets by utilizing the specific Vandermonde
Prior work on English monolingual retrieval has shown that a cross-encoder trained using a large number of relevance judgments for query-document pairs can be used as a teacher to train more efficient, but similarly effective, dual-encoder student models. Applying a similar knowledge distillation approach to training an efficient dual-encoder model for Cross-Language Information Retrieval (CLIR), where queries and documents are in different languages, is challenging due to the lack of a sufficiently large training collection when the query and document languages differ. The state of the art for CLIR thus relies on translating queries, documents, or both from the large English MS MARCO training set, an approach called Translate-Train. This paper proposes an alternative, Translate-Distill, in which knowledge distillation from either a monolingual cross-encoder or a CLIR cross-encoder is used to train a dual-encoder CLIR student model. This richer design space enables the teacher model to perform inference in an optimized setting, while training the student model directly for CLIR. Trained models and artifacts are publicly available on Huggingface.