PyTorchKR Papers

최신 논문

총 1,461편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2112.15238

Studying the Interplay between Information Loss and Operation Loss in Representations for Classification

Information-theoretic measures have been widely adopted in the design of features for learning and decision problems. Inspired by this, we look at the relationship between i) a weak form of information loss in the Shannon sense and ii) the operation loss in the minimum probability of error (MPE) sense when considering a family of lossy continuous representations (features) of a continuous observation. We present several results that shed light on this interplay. Our first result offers a lower bound on a weak form of information loss as a function of its respective operation loss when adopting a discrete lossy representation (quantization) instead of the original raw observation. From this, our main result shows that a specific form of vanishing information loss (a weak notion of asymptotic informational sufficiency) implies a vanishing MPE loss (or asymptotic operational sufficiency) when considering a general family of lossy continuous representations. Our theoretical findings support the observation that the selection of feature representations that attempt to capture informational sufficiency is appropriate for learning, but this selection is a rather conservative design principle if the intended goal is achieving MPE in classification. Supporting this last point, and under some structural conditions, we show that it is possible to adopt an alternative notion of informational sufficiency (strictly weaker than pure sufficiency in the mutual information sense) to achieve operational sufficiency in learning.

arXiv:2112.13237공식 코드 · Papers with Code

CABACE: Injecting Character Sequence Information and Domain Knowledge for Enhanced Acronym and Long-Form Extraction

Acronyms and long-forms are commonly found in research documents, more so in documents from scientific and legal domains. Many acronyms used in such documents are domain-specific and are very rarely found in normal text corpora. Owing to this, transformer-based NLP models often detect OOV (Out of Vocabulary) for acronym tokens, especially for non-English languages, and their performance suffers while linking acronyms to their long forms during extraction. Moreover, pretrained transformer models like BERT are not specialized to handle scientific and legal documents. With these points being the overarching motivation behind this work, we propose a novel framework CABACE: Character-Aware BERT for ACronym Extraction, which takes into account character sequences in text and is adapted to scientific and legal domains by masked language modelling. We further use an objective with an augmented loss function, adding the max loss and mask loss terms to the standard cross-entropy loss for training CABACE. We further leverage pseudo labelling and adversarial data generation to improve the generalizability of the framework. Experimental results prove the superiority of the proposed framework in comparison to various baselines. Additionally, we show that the proposed framework is better suited than baseline models for zero-shot generalization to non-English languages, thus reinforcing the effectiveness of our approach. Our team BacKGProp secured the highest scores on the French dataset, second-highest on Danish and Vietnamese, and third-highest in the English-Legal dataset on the global leaderboard for the acronym extraction (AE) shared task at SDU AAAI-22.

arXiv:2112.07820공식 코드 · Papers with Code

Value Retrieval with Arbitrary Queries for Form-like Documents

We propose value retrieval with arbitrary queries for form-like documents to reduce human effort of processing forms. Unlike previous methods that only address a fixed set of field items, our method predicts target value for an arbitrary query based on the understanding of the layout and semantics of a form. To further boost model performance, we propose a simple document language modeling (SimpleDLM) strategy to improve document understanding on large-scale model pre-training. Experimental results show that our method outperforms previous designs significantly and the SimpleDLM further improves our performance on value retrieval by around 17% F1 score compared with the state-of-the-art pre-training method. Code is available at https://github.com/salesforce/QVR-SimpleDLM.

arXiv:2112.07443공식 코드 · Papers with Code

Text Classification Models for Form Entity Linking

Forms are a widespread type of template-based document used in a great variety of fields including, among others, administration, medicine, finance, or insurance. The automatic extraction of the information included in these documents is greatly demanded due to the increasing volume of forms that are generated in a daily basis. However, this is not a straightforward task when working with scanned forms because of the great diversity of templates with different location of form entities, and the quality of the scanned documents. In this context, there is a feature that is shared by all forms: they contain a collection of interlinked entities built as key-value (or label-value) pairs, together with other entities such as headers or images. In this work, we have tacked the problem of entity linking in forms by combining image processing techniques and a text classification model based on the BERT architecture. This approach achieves state-of-the-art results with a F1-score of 0.80 on the FUNSD dataset, a 5% improvement regarding the best previous method. The code of this project is available at https://github.com/mavillot/FUNSD-Entity-Linking.

arXiv:2112.06240

Improving Logical-Level Natural Language Generation with Topic-Conditioned Data Augmentation and Logical Form Generation

Logical Natural Language Generation, i.e., generating textual descriptions that can be logically entailed by a structured table, has been a challenge due to the low fidelity of the generation. \citet{chen2020logic2text} have addressed this problem by annotating interim logical programs to control the generation contents and semantics, and presented the task of table-aware logical form to text (Logic2text) generation. However, although table instances are abundant in the real world, logical forms paired with textual descriptions require costly human annotation work, which limits the performance of neural models. To mitigate this, we propose topic-conditioned data augmentation (TopicDA), which utilizes GPT-2 to generate unpaired logical forms and textual descriptions directly from tables. We further introduce logical form generation (LG), a dual task of Logic2text that requires generating a valid logical form based on a text description of a table. We also propose a semi-supervised learning approach to jointly train a Logic2text and an LG model with both labeled and augmented data. The two models benefit from each other by providing extra supervision signals through back-translation. Experimental results on the Logic2text dataset and the LG task demonstrate that our approach can effectively utilize the augmented data and outperform supervised baselines by a substantial margin.

arXiv:2112.05863

Directed Speech Separation for Automatic Speech Recognition of Long Form Conversational Speech

Many of the recent advances in speech separation are primarily aimed at synthetic mixtures of short audio utterances with high degrees of overlap. Most of these approaches need an additional stitching step to stitch the separated speech chunks for long form audio. Since most of the approaches involve Permutation Invariant training (PIT), the order of separated speech chunks is nondeterministic and leads to difficulty in accurately stitching homogenous speaker chunks for downstream tasks like Automatic Speech Recognition (ASR). Also, most of these models are trained with synthetic mixtures and do not generalize to real conversational data. In this paper, we propose a speaker conditioned separator trained on speaker embeddings extracted directly from the mixed signal using an over-clustering based approach. This model naturally regulates the order of the separated chunks without the need for an additional stitching step. We also introduce a data sampling strategy with real and synthetic mixtures which generalizes well to real conversation speech. With this model and data sampling technique, we show significant improvements in speaker-attributed word error rate (SA-WER) on Hub5 data.

arXiv:2112.06763

A Homotopy Algorithm for Optimal Transport

The optimal transport problem has many applications in machine learning, physics, biology, economics, etc. Although its goal is very clear and mathematically well-defined, finding its optimal solution can be challenging for large datasets in high-dimensional space. Here, we propose a homotopy algorithm that first transforms the problem into an easy form, by changing the target distribution. It then transforms the problem back to the original form through a series of iterations, tracing a path of solutions until it finds the optimal solution for the original problem. We define the homotopy path as a subspace rotation based on the orthogonal Procrustes problem, and then we discretize the homotopy path using eigenvalue decomposition of the rotation matrix. Our goal is to provide an algorithm with complexity bound $\mathcal{O}(n^2 \log(n))$, faster than the existing methods in the literature.

arXiv:0711.3675

Derivations of Normalized Mutual Information in Binary Classifications

This correspondence studies the basic problem of classifications - how to evaluate different classifiers. Although the conventional performance indexes, such as accuracy, are commonly used in classifier selection or evaluation, information-based criteria, such as mutual information, are becoming popular in feature/model selections. In this work, we propose to assess classifiers in terms of normalized mutual information (NI), which is novel and well defined in a compact range for classifier evaluation. We derive close-form relations of normalized mutual information with respect to accuracy, precision, and recall in binary classifications. By exploring the relations among them, we reveal that NI is actually a set of nonlinear functions, with a concordant power-exponent form, to each performance index. The relations can also be expressed with respect to precision and recall, or to false alarm and hitting rate (recall).

arXiv:2112.04355

COSMIC: fast closed-form identification from large-scale data for LTV systems

We introduce a closed-form method for identification of discrete-time linear time-variant systems from data, formulating the learning problem as a regularized least squares problem where the regularizer favors smooth solutions within a trajectory. We develop a closed-form algorithm with guarantees of optimality and with a complexity that increases linearly with the number of instants considered per trajectory. The COSMIC algorithm achieves the desired result even in the presence of large volumes of data. Our method solved the problem using two orders of magnitude less computational power than a general purpose convex solver and was about 3 times faster than a Stochastic Block Coordinate Descent especially designed method. Computational times of our method remained in the order of magnitude of the second even for 10k and 100k time instants, where the general purpose solver crashed. To prove its applicability to real world systems, we test with spring-mass-damper system and use the estimated model to find the optimal control path. Our algorithm was applied to both a Low Fidelity and Functional Engineering Simulators for the Comet Interceptor mission, that requires precise pointing of the on-board cameras in a fast dynamics environment. Thus, this paper provides a fast alternative to classical system identification techniques for linear time-variant systems, while proving to be a solid base for applications in the Space industry and a step forward to the incorporation of algorithms that leverage data in such a safety-critical environment.

arXiv:2112.00894

Context-Dependent Semantic Parsing for Temporal Relation Extraction

Extracting temporal relations among events from unstructured text has extensive applications, such as temporal reasoning and question answering. While it is difficult, recent development of Neural-symbolic methods has shown promising results on solving similar tasks. Current temporal relation extraction methods usually suffer from limited expressivity and inconsistent relation inference. For example, in TimeML annotations, the concept of intersection is absent. Additionally, current methods do not guarantee the consistency among the predicted annotations. In this work, we propose SMARTER, a neural semantic parser, to extract temporal information in text effectively. SMARTER parses natural language to an executable logical form representation, based on a custom typed lambda calculus. In the training phase, dynamic programming on denotations (DPD) technique is used to provide weak supervision on logical forms. In the inference phase, SMARTER generates a temporal relation graph by executing the logical form. As a result, our neural semantic parser produces logical forms capturing the temporal information of text precisely. The accurate logical form representations of an event given the context ensure the correctness of the extracted relations.

arXiv:2112.00941공식 코드 · Papers with Code

Generalized Closed-form Formulae for Feature-based Subpixel Alignment in Patch-based Matching

Cost-based image patch matching is at the core of various techniques in computer vision, photogrammetry and remote sensing. When the subpixel disparity between the reference patch in the source and target images is required, either the cost function or the target image have to be interpolated. While cost-based interpolation is the easiest to implement, multiple works have shown that image based interpolation can increase the accuracy of the subpixel matching, but usually at the cost of expensive search procedures. This, however, is problematic, especially for very computation intensive applications such as stereo matching or optical flow computation. In this paper, we show that closed form formulae for subpixel disparity computation for the case of one dimensional matching, e.g., in the case of rectified stereo images where the search space is of one dimension, exists when using the standard NCC, SSD and SAD cost functions. We then demonstrate how to generalize the proposed formulae to the case of high dimensional search spaces, which is required for unrectified stereo matching and optical flow extraction. We also compare our results with traditional cost volume interpolation formulae as well as with state-of-the-art cost-based refinement methods, and show that the proposed formulae bring a small improvement over the state-of-the-art cost-based methods in the case of one dimensional search spaces, and a significant improvement when the search space is two dimensional.

arXiv:2111.14158

Generalized Fully Coherent Closed-form Receiver Design for Joint Radar and Communication System

In conventional radar, the transmission of the same waveform is repeated after a predefined interval of time called pulse-repetition-interval (PRI). This technique helps to estimate the range and Doppler shift of targets and suppress clutter. In dual-function radar communication (DFRC), different waveforms are transmitted after each PRI. Thus, each waveform yields different range-side-lobe (RSL) levels at the receiver's output. As a consequence, Doppler shift estimation and clutter suppression become challenging tasks. A state-of-the-art (SOTA) method claims that if the number of waveforms is more than two, it is impossible to achieve fully coherent RSL levels with both waveforms. Therefore, this algorithm uses iterative methods to achieve as much as possible coherency and minimize the RSL levels. In contrast to that SOTA method, we proposed two novel closed-form receivers for the DFRC that yield a fully coherent response for several waveforms and suppress the RSL levels. Experimental results demonstrate that the proposed receivers achieve full coherency and the RSL levels are significantly lower than the conventional method.

arXiv:2111.10858

Bilevel learning of l1-regularizers with closed-form gradients(BLORC)

We present a method for supervised learning of sparsity-promoting regularizers, a key ingredient in many modern signal reconstruction problems. The parameters of the regularizer are learned to minimize the mean squared error of reconstruction on a training set of ground truth signal and measurement pairs. Training involves solving a challenging bilevel optimization problem with a nonsmooth lower-level objective. We derive an expression for the gradient of the training loss using the implicit closed-form solution of the lower-level variational problem given by its dual problem, and provide an accompanying gradient descent algorithm (dubbed BLORC) to minimize the loss. Our experiments on simple natural images and for denoising 1D signals show that the proposed method can learn meaningful operators and the analytical gradients calculated are faster than standard automatic differentiation methods. While the approach we present is applied to denoising, we believe that it can be adapted to a wide-variety of inverse problems with linear measurement models, thus giving it applicability in a wide range of scenarios.

arXiv:2111.10692

Textbook to triples: Creating knowledge graph in the form of triples from AI TextBook

A knowledge graph is an essential and trending technology with great applications in entity recognition, search, or question answering. There are a plethora of methods in natural language processing for performing the task of Named entity recognition; however, there are very few methods that could provide triples for a domain-specific text. In this paper, an effort has been made towards developing a system that could convert the text from a given textbook into triples that can be used to visualize as a knowledge graph and use for further applications. The initial assessment and evaluation gave promising results with an F1 score of 82%.

arXiv:2111.11249

LeQua@CLEF2022: Learning to Quantify

LeQua 2022 is a new lab for the evaluation of methods for "learning to quantify" in textual datasets, i.e., for training predictors of the relative frequencies of the classes of interest in sets of unlabelled textual documents. While these predictions could be easily achieved by first classifying all documents via a text classifier and then counting the numbers of documents assigned to the classes, a growing body of literature has shown this approach to be suboptimal, and has proposed better methods. The goal of this lab is to provide a setting for the comparative evaluation of methods for learning to quantify, both in the binary setting and in the single-label multiclass setting. For each such setting we provide data either in ready-made vector form or in raw document form.

arXiv:2111.09191공식 코드 · Papers with Code

Preference Communication in Multi-Objective Normal-Form Games

We consider preference communication in two-player multi-objective normal-form games. In such games, the payoffs resulting from joint actions are vector-valued. Taking a utility-based approach, we assume there exists a utility function for each player which maps vectors to scalar utilities and consider agents that aim to maximise the utility of expected payoff vectors. As agents typically do not know their opponent's utility function or strategy, they must learn policies to interact with each other. Inspired by Stackelberg games, we introduce four novel preference communication protocols to aid agents in arriving at adequate solutions. Each protocol describes a specific approach for one agent to communicate preferences over their actions and how another agent responds. Additionally, to study when communication emerges, we introduce a communication protocol where agents must learn when to communicate. These protocols are subsequently evaluated on a set of five benchmark games against baseline agents that do not communicate. We find that preference communication can alter the learning process and lead to the emergence of cyclic policies which had not been previously observed in this setting. We further observe that the resulting policies can heavily depend on the characteristics of the game that is played. Lastly, we find that communication naturally emerges in both cooperative and self-interested settings.

arXiv:2111.08227

Phase function estimation from a diffuse optical image via deep learning

The phase function is a key element of a light propagation model for Monte Carlo (MC) simulation, which is usually fitted with an analytic function with associated parameters. In recent years, machine learning methods were reported to estimate the parameters of the phase function of a particular form such as the Henyey-Greenstein phase function but, to our knowledge, no studies have been performed to determine the form of the phase function. Here we design a convolutional neural network to estimate the phase function from a diffuse optical image without any explicit assumption on the form of the phase function. Specifically, we use a Gaussian mixture model as an example to represent the phase function generally and learn the model parameters accurately. The Gaussian mixture model is selected because it provides the analytic expression of phase function to facilitate deflection angle sampling in MC simulation, and does not significantly increase the number of free parameters. Our proposed method is validated on MC-simulated reflectance images of typical biological tissues using the Henyey-Greenstein phase function with different anisotropy factors. The effects of field of view (FOV) and spatial resolution on the errors are analyzed to optimize the estimation method. The mean squared error of the phase function is 0.01 and the relative error of the anisotropy factor is 3.28%.

arXiv:2111.08267

Solving Probability and Statistics Problems by Program Synthesis

We solve university level probability and statistics questions by program synthesis using OpenAI's Codex, a Transformer trained on text and fine-tuned on code. We transform course problems from MIT's 18.05 Introduction to Probability and Statistics and Harvard's STAT110 Probability into programming tasks. We then execute the generated code to get a solution. Since these course questions are grounded in probability, we often aim to have Codex generate probabilistic programs that simulate a large number of probabilistic dependencies to compute its solution. Our approach requires prompt engineering to transform the question from its original form to an explicit, tractable form that results in a correct program and solution. To estimate the amount of work needed to translate an original question into its tractable form, we measure the similarity between original and transformed questions. Our work is the first to introduce a new dataset of university-level probability and statistics problems and solve these problems in a scalable fashion using the program synthesis capabilities of large language models.

arXiv:2111.08102

The Partially Observable History Process

We introduce the partially observable history process (POHP) formalism for reinforcement learning. POHP centers around the actions and observations of a single agent and abstracts away the presence of other players without reducing them to stochastic processes. Our formalism provides a streamlined interface for designing algorithms that defy categorization as exclusively single or multi-agent, and for developing theory that applies across these domains. We show how the POHP formalism unifies traditional models including the Markov decision process, the Markov game, the extensive-form game, and their partially observable extensions, without introducing burdensome technical machinery or violating the philosophical underpinnings of reinforcement learning. We illustrate the utility of our formalism by concisely exploring observable sequential rationality, examining some theoretical properties of general immediate regret minimization, and generalizing the extensive-form regret minimization (EFR) algorithm.

arXiv:2111.07019

Closed-form Two-way TOA Localization and Synchronization for User Devices with Motion and Clock Drift

A two-way time-of-arrival (TOA system is composed of anchor nodes (ANs and user devices (UDs . Two-way TOA measurements between AN-UD pairs are obtained via round-trip communications to achieve localization and synchronization (LAS for a UD. Existing LAS method for a moving UD with clock drift adopts an iterative algorithm, which requires accurate initialization and has high computational complexity. In this paper, we propose a new closed-form two-way TOA LAS approach, namely CFTWLAS, which does not require initialization, has low complexity and empirically achieves optimal LAS accuracy. We first linearize the LAS problem by squaring and differencing the two-way TOA equations. We employ two auxiliary variables to simplify the problem to finding the analytical solution of quadratic equations. Due to the measurement noise, we can only obtain a raw LAS estimation from the solution of the auxiliary variables. Then, a weighted least squares step is applied to further refine the raw estimation. We analyze the theoretical error of the new CFTWLAS and show that it empirically reaches the Cramer-Rao lower bound (CRLB with sufficient ANs under the condition with proper geometry and small noise. Numerical results in a 3D scenario verify the theoretical analysis that the estimation accuracy of the new CFTWLAS method reaches CRLB in the presented experiments when the number of the ANs is large, the geometry is appropriate, and the noise is small. Unlike the iterative method whose complexity increases with the iteration count, the new CFTWLAS has constant low complexity.