PyTorchKR Papers

최신 논문

총 1,461편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2305.09957

Quantum neural networks form Gaussian processes

It is well known that artificial neural networks initialized from independent and identically distributed priors converge to Gaussian processes in the limit of a large number of neurons per hidden layer. In this work we prove an analogous result for Quantum Neural Networks (QNNs). Namely, we show that the outputs of certain models based on Haar random unitary or orthogonal deep QNNs converge to Gaussian processes in the limit of large Hilbert space dimension $d$. The derivation of this result is more nuanced than in the classical case due to the role played by the input states, the measurement observable, and the fact that the entries of unitary matrices are not independent. Then, we show that the efficiency of predicting measurements at the output of a QNN using Gaussian process regression depends on the observable's bodyness. Furthermore, our theorems imply that the concentration of measure phenomenon in Haar random QNNs is worse than previously thought, as we prove that expectation values and gradients concentrate as $\mathcal{O}\left(\frac{1}{e^d \sqrt{d}}\right)$. Finally, we discuss how our results improve our understanding of concentration in $t$-designs.

arXiv:2305.09354

Control of distributed-parameter systems using normal forms: An introduction

This paper gives an overview of the control of distributed-parameter systems using normal forms. Considering linear controllable PDE-ODE systems of hyperbolic type, two methods derive tracking controllers by mapping the system into a form that is advantageous for the control design, analogous to the finite-dimensional case. A flatness-based controller makes use of the hyperbolic controller canonical form that follows from a parametrization of the system's solutions. A backstepping design exploits the strict-feedback form of the system to recursively stabilize and transform the subsystems.

arXiv:2305.07615공식 코드 · Papers with Code

What are the Desired Characteristics of Calibration Sets? Identifying Correlates on Long Form Scientific Summarization

Summarization models often generate text that is poorly calibrated to quality metrics because they are trained to maximize the likelihood of a single reference (MLE). To address this, recent work has added a calibration step, which exposes a model to its own ranked outputs to improve relevance or, in a separate line of work, contrasts positive and negative sets to improve faithfulness. While effective, much of this work has focused on how to generate and optimize these sets. Less is known about why one setup is more effective than another. In this work, we uncover the underlying characteristics of effective sets. For each training instance, we form a large, diverse pool of candidates and systematically vary the subsets used for calibration fine-tuning. Each selection strategy targets distinct aspects of the sets, such as lexical diversity or the size of the gap between positive and negatives. On three diverse scientific long-form summarization datasets (spanning biomedical, clinical, and chemical domains), we find, among others, that faithfulness calibration is optimal when the negative sets are extractive and more likely to be generated, whereas for relevance calibration, the metric margin between candidates should be maximized and surprise--the disagreement between model and metric defined candidate rankings--minimized. Code to create, select, and optimize calibration sets is available at https://github.com/griff4692/calibrating-summaries

arXiv:2305.07500공식 코드 · Papers with Code

Learning representations that are closed-form Monge mapping optimal with application to domain adaptation

Optimal transport (OT) is a powerful geometric tool used to compare and align probability measures following the least effort principle. Despite its widespread use in machine learning (ML), OT problem still bears its computational burden, while at the same time suffering from the curse of dimensionality for measures supported on general high-dimensional spaces. In this paper, we propose to tackle these challenges using representation learning. In particular, we seek to learn an embedding space such that the samples of the two input measures become alignable in it with a simple affine mapping that can be calculated efficiently in closed-form. We then show that such approach leads to results that are comparable to solving the original OT problem when applied to the transfer learning task on which many OT baselines where previously evaluated in both homogeneous and heterogeneous DA settings. The code for our contribution is available at \url{https://github.com/Oleffa/LaOT}.

arXiv:2305.06849공식 코드 · Papers with Code

WebCPM: Interactive Web Search for Chinese Long-form Question Answering

Long-form question answering (LFQA) aims at answering complex, open-ended questions with detailed, paragraph-length responses. The de facto paradigm of LFQA necessitates two procedures: information retrieval, which searches for relevant supporting facts, and information synthesis, which integrates these facts into a coherent answer. In this paper, we introduce WebCPM, the first Chinese LFQA dataset. One unique feature of WebCPM is that its information retrieval is based on interactive web search, which engages with a search engine in real time. Following WebGPT, we develop a web search interface. We recruit annotators to search for relevant information using our interface and then answer questions. Meanwhile, the web search behaviors of our annotators would be recorded. In total, we collect 5,500 high-quality question-answer pairs, together with 14,315 supporting facts and 121,330 web search actions. We fine-tune pre-trained language models to imitate human behaviors for web search and to generate answers based on the collected facts. Our LFQA pipeline, built on these fine-tuned models, generates answers that are no worse than human-written ones in 32.5% and 47.5% of the cases on our dataset and DuReader, respectively.

arXiv:2305.04460공식 코드 · Papers with Code

Language Independent Neuro-Symbolic Semantic Parsing for Form Understanding

Recent works on form understanding mostly employ multimodal transformers or large-scale pre-trained language models. These models need ample data for pre-training. In contrast, humans can usually identify key-value pairings from a form only by looking at layouts, even if they don't comprehend the language used. No prior research has been conducted to investigate how helpful layout information alone is for form understanding. Hence, we propose a unique entity-relation graph parsing method for scanned forms called LAGNN, a language-independent Graph Neural Network model. Our model parses a form into a word-relation graph in order to identify entities and relations jointly and reduce the time complexity of inference. This graph is then transformed by deterministic rules into a fully connected entity-relation graph. Our model simply takes into account relative spacing between bounding boxes from layout information to facilitate easy transfer across languages. To further improve the performance of LAGNN, and achieve isomorphism between entity-relation graphs and word-relation graphs, we use integer linear programming (ILP) based inference. Code is publicly available at https://github.com/Bhanu068/LAGNN

arXiv:2305.02549

FormNetV2: Multimodal Graph Contrastive Learning for Form Document Information Extraction

The recent advent of self-supervised pre-training techniques has led to a surge in the use of multimodal learning in form document understanding. However, existing approaches that extend the mask language modeling to other modalities require careful multi-task tuning, complex reconstruction target designs, or additional pre-training data. In FormNetV2, we introduce a centralized multimodal graph contrastive learning strategy to unify self-supervised pre-training for all modalities in one loss. The graph contrastive objective maximizes the agreement of multimodal representations, providing a natural interplay for all modalities without special customization. In addition, we extract image features within the bounding box that joins a pair of tokens connected by a graph edge, capturing more targeted visual cues without loading a sophisticated and separately pre-trained image embedder. FormNetV2 establishes new state-of-the-art performance on FUNSD, CORD, SROIE and Payment benchmarks with a more compact model size.

arXiv:2305.01903

Input-Output Feedback Linearization Preserving Task Priority for Multivariate Nonlinear Systems Having Singular Input Gain Matrix

We propose an extension of the input-output feedback linearization for a class of multivariate systems that are not input-output linearizable in a classical manner. The key observation is that the usual input-output linearization problem can be interpreted as the problem of solving simultaneous linear equations associated with the input gain matrix: thus, even at points where the input gain matrix becomes singular, it is still possible to solve a part of linear equations, by which a subset of input-output relations is made linear or close to be linear. Based on this observation, we adopt the task priority-based approach in the input-output linearization problem. First, we generalize the classical Byrnes-Isidori normal form to a prioritized normal form having a triangular structure, so that the singularity of a subblock of the input gain matrix related to lower-priority tasks does not directly propagate to higher-priority tasks. Next, we present a prioritized input-output linearization via the multi-objective optimization with the lexicographical ordering, resulting in a prioritized semilinear form that establishes input output relations whose subset with higher priority is linear or close to be linear. Finally, Lyapunov analysis on ultimate boundedness and task achievement is provided, particularly when the proposed prioritized input-output linearization is applied to the output tracking problem. This work introduces a new control framework for complex systems having critical and noncritical control issues, by assigning higher priority to the critical ones.

arXiv:2305.01965공식 코드 · Papers with Code

Analysing the Impact of Audio Quality on the Use of Naturalistic Long-Form Recordings for Infant-Directed Speech Research

Modelling of early language acquisition aims to understand how infants bootstrap their language skills. The modelling encompasses properties of the input data used for training the models, the cognitive hypotheses and their algorithmic implementations being tested, and the evaluation methodologies to compare models to human data. Recent developments have enabled the use of more naturalistic training data for computational models. This also motivates development of more naturalistic tests of model behaviour. A crucial step towards such an aim is to develop representative speech datasets consisting of speech heard by infants in their natural environments. However, a major drawback of such recordings is that they are typically noisy, and it is currently unclear how the sound quality could affect analyses and modelling experiments conducted on such data. In this paper, we explore this aspect for the case of infant-directed speech (IDS) and adult-directed speech (ADS) analysis. First, we manually and automatically annotated audio quality of utterances extracted from two corpora of child-centred long-form recordings (in English and French). We then compared acoustic features of IDS and ADS in an in-lab dataset and across different audio quality subsets of naturalistic data. Finally, we assessed how the audio quality and recording environment may change the conclusions of a modelling analysis using a recent self-supervised learning model. Our results show that the use of modest and high audio quality naturalistic speech data result in largely similar conclusions on IDS and ADS in terms of acoustic analyses and modelling experiments. We also found that an automatic sound quality assessment tool can be used to screen out useful parts of long-form recordings for a closer analysis with comparable results to that of manual quality annotation.

arXiv:2304.14956

PAO: A general particle swarm algorithm with exact dynamics and closed-form transition densities

A great deal of research has been conducted in the consideration of meta-heuristic optimisation methods that are able to find global optima in settings that gradient based optimisers have traditionally struggled. Of these, so-called particle swarm optimisation (PSO) approaches have proven to be highly effective in a number of application areas. Given the maturity of the PSO field, it is likely that novel variants of the PSO algorithm stand to offer only marginal gains in terms of performance -- there is, after all, no free lunch. Instead of only chasing performance on suites of benchmark optimisation functions, it is argued herein that research effort is better placed in the pursuit of algorithms that also have other useful properties. In this work, a highly-general, interpretable variant of the PSO algorithm -- particle attractor algorithm (PAO) -- is proposed. Furthermore, the algorithm is designed such that the transition densities (describing the motions of the particles from one generation to the next) can be computed exactly in closed form for each step. Access to closed-form transition densities has important ramifications for the closely-related field of Sequential Monte Carlo (SMC). In order to demonstrate that the useful properties do not come at the cost of performance, PAO is compared to several other state-of-the art heuristic optimisation algorithms in a benchmark comparison study.

arXiv:2304.13866

Optimal tie-breaking rules

We consider two-player contests with the possibility of ties and study the effect of different tie-breaking rules on effort. For ratio-form and difference-form contests that admit pure-strategy Nash equilibrium, we find that the effort of both players is monotone decreasing in the probability that ties are broken in favor of the stronger player. Thus, the effort-maximizing tie-breaking rule commits to breaking ties in favor of the weaker agent. With symmetric agents, we find that the equilibrium is generally symmetric and independent of the tie-breaking rule. We also study the design of random tie-breaking rules that are ex-ante fair and identify sufficient conditions under which breaking ties before the contest actually leads to greater expected effort than the more commonly observed practice of breaking ties after the contest.

arXiv:2304.11974

A Closed-form Expression for the Gaussian Noise Model in the Presence of Raman Amplification

A closed-form model for the nonlinear interference (NLI) in Raman amplified links is presented, the formula accounts for both forward (FW) and backward (BW) pumping schemes and inter-channel stimulated Raman scattering (ISRS) effect. The formula also accounts for an arbitrary number of pumps, wavelength-dependent fibre parameters, launch-power profiles, and is tested over a distributed Raman-amplified system setup. The formula is suitable for ultra-wideband (UWB) optical transmission systems and is applied in a signal with 13~THz optical bandwidth corresponding to transmission over the S-, C-, and L- band. The accuracy of the closed-form formula is validated through comparison with numerical integration of the Gaussian noise (GN) model and split-step Fourier method (SSFM) simulations in a point-to-point transmission link.

arXiv:2304.11010

Invariance properties of maximal extractable value

We develop a formalism for reasoning about trading on decentralized exchanges on blockchains and a formulation of a particular form of maximal extractable value (MEV) that represents the total arbitrage opportunity extractable from on-chain liquidity. We use this formalism to prove that for blockchains with deterministic block times whose liquidity pools satisfy some natural properties that are satisfied by pools in practice, this form of MEV is invariant under changes to the ordering mechanism of the blockchain and distribution of block times. We do this by characterizing the MEV as the profit of a particularly simple arbitrage strategy when left uncontested. These results can inform design of blockchain protocols by ruling out designs aiming to increase trading opportunity by changing the ordering mechanism or shortening block times.

arXiv:2304.07957

A Question-Answering Approach to Key Value Pair Extraction from Form-like Document Images

In this paper, we present a new question-answering (QA) based key-value pair extraction approach, called KVPFormer, to robustly extracting key-value relationships between entities from form-like document images. Specifically, KVPFormer first identifies key entities from all entities in an image with a Transformer encoder, then takes these key entities as \textbf{questions} and feeds them into a Transformer decoder to predict their corresponding \textbf{answers} (i.e., value entities) in parallel. To achieve higher answer prediction accuracy, we propose a coarse-to-fine answer prediction approach further, which first extracts multiple answer candidates for each identified question in the coarse stage and then selects the most likely one among these candidates in the fine stage. In this way, the learning difficulty of answer prediction can be effectively reduced so that the prediction accuracy can be improved. Moreover, we introduce a spatial compatibility attention bias into the self-attention/cross-attention mechanism for \Ours{} to better model the spatial interactions between entities. With these new techniques, our proposed \Ours{} achieves state-of-the-art results on FUNSD and XFUND datasets, outperforming the previous best-performing method by 7.2\% and 13.2\% in F1 score, respectively.

arXiv:2304.07838

Discrete-Time State-Feedback Controller with Canonical Form on Inverted Pendulum (on a cart)

The scope of inverted pendulum has been widely studied as one of the notable research with respect to standing in balance. The concept of this pendulum is similar to missile guidance, meaning that the center of drag is ahead that of gravity. Mathematical model of inverted pendulum on a cart is moreover presented in this paper. Various rewarding parameters are proposed from the displacement of the pivot, angular rotation, to external force exerted on the carriage so as to gain its equilibrium points and the linearized systems. Due to the severe risk of instability, a reliable closed-loop state feedback controller is designed to stabilize in upright position, even with large deviations. The specific concept proposed is to apply the canonical form of computing the determinant of gain $K$ leading to $K_d$. The results show that the constructed design can maintain the stability of the system by applying three sorts of initial condition and choosing sampling time $T$ under $0.2$ with small possible degrading performance.

arXiv:2304.02560

VicTR: Video-conditioned Text Representations for Activity Recognition

Vision-Language models (VLMs) have excelled in the image-domain -- especially in zero-shot settings -- thanks to the availability of vast pretraining data (i.e., paired image-text samples). However for videos, such paired data is not as abundant. Therefore, video-VLMs are usually designed by adapting pretrained image-VLMs to the video-domain, instead of training from scratch. All such recipes rely on augmenting visual embeddings with temporal information (i.e., image $\rightarrow$ video), often keeping text embeddings unchanged or even being discarded. In this paper, we argue the contrary, that better video-VLMs can be designed by focusing more on augmenting text, rather than visual information. More specifically, we introduce Video-conditioned Text Representations (VicTR): a form of text embeddings optimized w.r.t. visual embeddings, creating a more-flexible contrastive latent space. Our model can further make use of freely-available semantic information, in the form of visually-grounded auxiliary text (e.g. object or scene information). We evaluate our model on few-shot, zero-shot (HMDB-51, UCF-101), short-form (Kinetics-400) and long-form (Charades) activity recognition benchmarks, showing strong performance among video-VLMs.

arXiv:2304.01859

Unified Behavioral Data-Driven Performance Analysis: A Generalized Plant Approach

In this paper, we present a novel approach to combine data-driven non-parametric representations with model-based representations of dynamical systems. Based on a data-driven form of linear fractional transformations, we introduce a data-driven form of generalized plants. This form can be leveraged to accomplish performance characterizations, e.g., in the form of a mixed-sensitivity approach, and LMI-based conditions to verify finite-horizon dissipativity. In particular, we show how finite-horizon $\ell_2$-gain under weighting filter-based general performance specifications are verified for implemented controllers on systems for which only input-output data is available. The overall effectiveness of the proposed method is demonstrated by simulation examples.

arXiv:2304.01577공식 코드 · Papers with Code

Form-NLU: Dataset for the Form Natural Language Understanding

Compared to general document analysis tasks, form document structure understanding and retrieval are challenging. Form documents are typically made by two types of authors; A form designer, who develops the form structure and keys, and a form user, who fills out form values based on the provided keys. Hence, the form values may not be aligned with the form designer's intention (structure and keys) if a form user gets confused. In this paper, we introduce Form-NLU, the first novel dataset for form structure understanding and its key and value information extraction, interpreting the form designer's intent and the alignment of user-written value on it. It consists of 857 form images, 6k form keys and values, and 4k table keys and values. Our dataset also includes three form types: digital, printed, and handwritten, which cover diverse form appearances and layouts. We propose a robust positional and logical relation-based form key-value information extraction framework. Using this dataset, Form-NLU, we first examine strong object detection models for the form layout understanding, then evaluate the key information extraction task on the dataset, providing fine-grained results for different types of forms and keys. Furthermore, we examine it with the off-the-shelf pdf layout extraction tool and prove its feasibility in real-world cases.

arXiv:2303.14526

Selective Structured State-Spaces for Long-Form Video Understanding

Effective modeling of complex spatiotemporal dependencies in long-form videos remains an open problem. The recently proposed Structured State-Space Sequence (S4) model with its linear complexity offers a promising direction in this space. However, we demonstrate that treating all image-tokens equally as done by S4 model can adversely affect its efficiency and accuracy. To address this limitation, we present a novel Selective S4 (i.e., S5) model that employs a lightweight mask generator to adaptively select informative image tokens resulting in more efficient and accurate modeling of long-term spatiotemporal dependencies in videos. Unlike previous mask-based token reduction methods used in transformers, our S5 model avoids the dense self-attention calculation by making use of the guidance of the momentum-updated S4 model. This enables our model to efficiently discard less informative tokens and adapt to various long-form video understanding tasks more effectively. However, as is the case for most token reduction methods, the informative image tokens could be dropped incorrectly. To improve the robustness and the temporal horizon of our model, we propose a novel long-short masked contrastive learning (LSMCL) approach that enables our model to predict longer temporal context using shorter input videos. We present extensive comparative results using three challenging long-form video understanding datasets (LVU, COIN and Breakfast), demonstrating that our approach consistently outperforms the previous state-of-the-art S4 model by up to 9.6% accuracy while reducing its memory footprint by 23%.

arXiv:2303.13716공식 코드 · Papers with Code

ReCOGS: How Incidental Details of a Logical Form Overshadow an Evaluation of Semantic Interpretation

Compositional generalization benchmarks for semantic parsing seek to assess whether models can accurately compute meanings for novel sentences, but operationalize this in terms of logical form (LF) prediction. This raises the concern that semantically irrelevant details of the chosen LFs could shape model performance. We argue that this concern is realized for the COGS benchmark. COGS poses generalization splits that appear impossible for present-day models, which could be taken as an indictment of those models. However, we show that the negative results trace to incidental features of COGS LFs. Converting these LFs to semantically equivalent ones and factoring out capabilities unrelated to semantic interpretation, we find that even baseline models get traction. A recent variable-free translation of COGS LFs suggests similar conclusions, but we observe this format is not semantically equivalent; it is incapable of accurately representing some COGS meanings. These findings inform our proposal for ReCOGS, a modified version of COGS that comes closer to assessing the target semantic capabilities while remaining very challenging. Overall, our results reaffirm the importance of compositional generalization and careful benchmark task design.