arXiv subjects

Yitong Li

Publications and source records attributed to Yitong Li.

At least 109 records · Page 6Linked to original sources

arXiv2022

Polarimetric Pose Prediction

Light has many properties that vision sensors can passively measure. Colour-band separated wavelength and intensity are arguably the most commonly used for monocular 6D object pose estimation. This paper explores how complementary polarisation information, i.e. the orientation of light wave oscillations, influences the accuracy of pose predictions. A hybrid model that leverages physical priors jointly with a data-driven learning strategy is designed and carefully tested on objects with different levels of photometric complexity. Our design significantly improves the pose accuracy compared to state-of-the-art photometric approaches and enables object pose estimation for highly reflective and transparent objects. A new multi-modal instance-level 6D object pose dataset with highly accurate pose annotations for multiple objects with varying photometric complexity is introduced as a benchmark.

cs.CV↗

arXiv2022

PanGu-Bot: Efficient Generative Dialogue Pre-training from Pre-trained Language Model

In this paper, we introduce PanGu-Bot, a Chinese pre-trained open-domain dialogue generation model based on a large pre-trained language model (PLM) PANGU-alpha (Zeng et al.,2021). Different from other pre-trained dialogue models trained over a massive amount of dialogue data from scratch, we aim to build a powerful dialogue model with relatively fewer data and computation costs by inheriting valuable language capabilities and knowledge from PLMs. To this end, we train PanGu-Bot from the large PLM PANGU-alpha, which has been proven well-performed on a variety of Chinese natural language tasks. We investigate different aspects of responses generated by PanGu-Bot, including response quality, knowledge, and safety. We show that PanGu-Bot outperforms state-of-the-art Chinese dialogue systems (CDIALGPT (Wang et al., 2020), EVA (Zhou et al., 2021), EVA2.0 (Gu et al., 2022)) w.r.t. the above three aspects. We also demonstrate that PanGu-Bot can be easily deployed to generate emotional responses without further training. Throughout our empirical analysis, we also point out that the PanGu-Bot response quality, knowledge correctness, and safety are still far from perfect, and further explorations are indispensable to building reliable and smart dialogue systems. Our model and code will be available at https://github.com/huawei-noah/Pretrained-Language-Model/tree/master/PanGu-Bot soon.

cs.CL↗

arXiv2022

PhoCaL: A Multi-Modal Dataset for Category-Level Object Pose Estimation with Photometrically Challenging Objects

Object pose estimation is crucial for robotic applications and augmented reality. Beyond instance level 6D object pose estimation methods, estimating category-level pose and shape has become a promising trend. As such, a new research field needs to be supported by well-designed datasets. To provide a benchmark with high-quality ground truth annotations to the community, we introduce a multimodal dataset for category-level object pose estimation with photometrically challenging objects termed PhoCaL. PhoCaL comprises 60 high quality 3D models of household objects over 8 categories including highly reflective, transparent and symmetric objects. We developed a novel robot-supported multi-modal (RGB, depth, polarisation) data acquisition and annotation process. It ensures sub-millimeter accuracy of the pose for opaque textured, shiny and transparent objects, no motion blur and perfect camera synchronisation. To set a benchmark for our dataset, state-of-the-art RGB-D and monocular RGB methods are evaluated on the challenging scenes of PhoCaL.

cs.CV↗

arXiv2022

Is my Depth Ground-Truth Good Enough? HAMMER -- Highly Accurate Multi-Modal Dataset for DEnse 3D Scene Regression

Depth estimation is a core task in 3D computer vision. Recent methods investigate the task of monocular depth trained with various depth sensor modalities. Every sensor has its advantages and drawbacks caused by the nature of estimates. In the literature, mostly mean average error of the depth is investigated and sensor capabilities are typically not discussed. Especially indoor environments, however, pose challenges for some devices. Textureless regions pose challenges for structure from motion, reflective materials are problematic for active sensing, and distances for translucent material are intricate to measure with existing sensors. This paper proposes HAMMER, a dataset comprising depth estimates from multiple commonly used sensors for indoor depth estimation, namely ToF, stereo, structured light together with monocular RGB+P data. We construct highly reliable ground truth depth maps with the help of 3D scanners and aligned renderings. A popular depth estimators is trained on this data and typical depth senosors. The estimates are extensively analyze on different scene structures. We notice generalization issues arising from various sensor technologies in household environments with challenging but everyday scene content. HAMMER, which we make publicly available, provides a reliable base to pave the way to targeted depth improvements and sensor fusion approaches.

cs.CV↗

arXiv2022

fairlib: A Unified Framework for Assessing and Improving Classification Fairness

This paper presents fairlib, an open-source framework for assessing and improving classification fairness. It provides a systematic framework for quickly reproducing existing baseline models, developing new methods, evaluating models with different metrics, and visualizing their results. Its modularity and extensibility enable the framework to be used for diverse types of inputs, including natural language, images, and audio. In detail, we implement 14 debiasing methods, including pre-processing, at-training-time, and post-processing approaches. The built-in metrics cover the most commonly used fairness criterion and can be further generalized and customized for fairness evaluation.

cs.LG↗

arXiv2022

Impedance-based Root-cause Analysis: Comparative Study of Impedance Models and Calculation of Eigenvalue Sensitivity

Impedance models of power systems are useful when state-space models of apparatus such as inverter-based resources (IBRs) have not been made available and instead only black-box impedance models are available. For tracing the root causes of poor damping and tuning modes of the system, the sensitivity of the modes to components and parameters are needed. The so-called critical admittance-eigenvalue sensitivity based on nodal admittance model has provided a partial solution but omits meaningful directional information. The alternative whole-system impedance model yields participation factors of shunt-connected apparatus with directional information that allows separate tuning for damping and frequency, yet do not cover series-connected components. This paper formalises the relationships between the two forms of impedance models and between the two forms of root-cause analysis. The calculation of system eigenvalue sensitivity in impedance models is further developed, which fills the gaps of previous research and establishes a complete theory of impedance-based root-cause analysis. The theoretical relationships and the tuning of parameters have been illustrated with a three-node passive network, a modified IEEE 14-bus network and a modified NETS-NYPS 68-bus network, showing that tools can be developed for tuning of IBR-rich power systems where only black-box impedance models are available.

eess.SY↗

arXiv2022

CINS: Comprehensive Instruction for Few-shot Learning in Task-oriented Dialog Systems

As labeling cost for different modules in task-oriented dialog (ToD) systems is high, a major challenge in practice is to learn different tasks with the least amount of labeled data. Recently, prompting methods over pre-trained language models (PLMs) have shown promising results for few-shot learning in ToD. To better utilize the power of PLMs, this paper proposes Comprehensive Instruction (CINS) that exploits PLMs with extra task-specific instructions. We design a schema (definition, constraint, prompt) of instructions and their customized realizations for three important downstream tasks in ToD, i.e. intent classification, dialog state tracking, and natural language generation. A sequence-to-sequence model (T5) is adopted to solve these three tasks in a unified framework. Extensive experiments are conducted on these ToD tasks in realistic few-shot learning scenarios with small validation data. Empirical results demonstrate that the proposed CINS approach consistently improves techniques that finetune PLMs with raw input or short prompts.

cs.CL↗

arXiv2022

RotateQVS: Representing Temporal Information as Rotations in Quaternion Vector Space for Temporal Knowledge Graph Completion

Temporal factors are tied to the growth of facts in realistic applications, such as the progress of diseases and the development of political situation, therefore, research on Temporal Knowledge Graph (TKG) attracks much attention. In TKG, relation patterns inherent with temporality are required to be studied for representation learning and reasoning across temporal facts. However, existing methods can hardly model temporal relation patterns, nor can capture the intrinsic connections between relations when evolving over time, lacking of interpretability. In this paper, we propose a novel temporal modeling method which represents temporal entities as Rotations in Quaternion Vector Space (RotateQVS) and relations as complex vectors in Hamilton's quaternion space. We demonstrate our method can model key patterns of relations in TKG, such as symmetry, asymmetry, inverse, and can further capture time-evolved relations by theory. Empirically, we show that our method can boost the performance of link prediction tasks over four temporal knowledge graph benchmarks.

cs.AI↗

arXiv2022

Compilable Neural Code Generation with Compiler Feedback

Automatically generating compilable programs with (or without) natural language descriptions has always been a touchstone problem for computational linguistics and automated software engineering. Existing deep-learning approaches model code generation as text generation, either constrained by grammar structures in decoder, or driven by pre-trained language models on large-scale code corpus (e.g., CodeGPT, PLBART, and CodeT5). However, few of them account for compilability of the generated programs. To improve compilability of the generated programs, this paper proposes COMPCODER, a three-stage pipeline utilizing compiler feedback for compilable code generation, including language model fine-tuning, compilability reinforcement, and compilability discrimination. Comprehensive experiments on two code generation tasks demonstrate the effectiveness of our proposed approach, improving the success rate of compilation from 44.18 to 89.18 in code completion on average and from 70.3 to 96.2 in text-to-code generation, respectively, when comparing with the state-of-the-art CodeGPT.

cs.CL↗

arXiv2022

Revisiting Grid-Forming and Grid-Following Inverters: A Duality Theory

Power electronic converters for integrating renewable energy resources into power systems can be divided into grid-forming and grid-following inverters. They possess certain similarities, but several important differences, which means that the relationship between them is quite subtle and sometimes obscure. In this article, a new perspective based on duality is proposed to create new insights. It successfully unifies the grid interfacing and synchronization characteristics of the two inverter types in a symmetric, elegant, and technology-neutral form. Analysis shows that the grid-forming and grid-following inverters are duals of each other in several ways including a) synchronization controllers: frequency droop control and phase-locked loop (PLL); b) grid-interfacing characteristics: current-following voltage-forming and voltage-following current-forming; c) swing characteristics: current-angle swing and voltage-angle swing; d) inner-loop controllers: output impedance shaping and output admittance shaping; and e) grid strength compatibility: strong-grid instability and weak-grid instability. The swing equations are also derived in dual form, which reveal the dynamic interaction between the grid strength, the synchronization controllers, and the inner-loop controllers. Insights are generated into cases of poor stability in both small-signal and transient/large-signal. The theoretical analysis and simulation results are used to illustrate cases for simple single-inverter-infinite-bus systems and a multi-inverter power network.

eess.SY↗

arXiv2022

Improving Disentangled Text Representation Learning with Information-Theoretic Guidance

Learning disentangled representations of natural language is essential for many NLP tasks, e.g., conditional text generation, style transfer, personalized dialogue systems, etc. Similar problems have been studied extensively for other forms of data, such as images and videos. However, the discrete nature of natural language makes the disentangling of textual representations more challenging (e.g., the manipulation over the data space cannot be easily achieved). Inspired by information theory, we propose a novel method that effectively manifests disentangled representations of text, without any supervision on semantics. A new mutual information upper bound is derived and leveraged to measure dependence between style and content. By minimizing this upper bound, the proposed method induces style and content embeddings into two independent low-dimensional spaces. Experiments on both conditional text generation and text-style transfer demonstrate the high quality of our disentangled representation in terms of content and style preservation.

cs.LG↗

arXiv2021

Estimating Uncertainty Intervals from Collaborating Networks

Effective decision making requires understanding the uncertainty inherent in a prediction. In regression, this uncertainty can be estimated by a variety of methods; however, many of these methods are laborious to tune, generate overconfident uncertainty intervals, or lack sharpness (give imprecise intervals). We address these challenges by proposing a novel method to capture predictive distributions in regression by defining two neural networks with two distinct loss functions. Specifically, one network approximates the cumulative distribution function, and the second network approximates its inverse. We refer to this method as Collaborating Networks (CN). Theoretical analysis demonstrates that a fixed point of the optimization is at the idealized solution, and that the method is asymptotically consistent to the ground truth distribution. Empirically, learning is straightforward and robust. We benchmark CN against several common approaches on two synthetic and six real-world datasets, including forecasting A1c values in diabetic patients from electronic health records, where uncertainty is critical. In the synthetic data, the proposed approach essentially matches ground truth. In the real-world datasets, CN improves results on many performance metrics, including log-likelihood estimates, mean absolute errors, coverage estimates, and prediction interval widths.

stat.ML↗

arXiv2021

UniDS: A Unified Dialogue System for Chit-Chat and Task-oriented Dialogues

With the advances in deep learning, tremendous progress has been made with chit-chat dialogue systems and task-oriented dialogue systems. However, these two systems are often tackled separately in current methods. To achieve more natural interaction with humans, a dialogue agent needs to be capable of both chatting and accomplishing tasks. To this end, we propose a unified dialogue system (UniDS) with the two aforementioned skills. In particular, we design a unified dialogue data schema, compatible for both chit-chat and task-oriented dialogues, and we train UniDS with mixed dialogue data from a pretrained chit-chat dialogue model. Without adding extra parameters to SOTA baselines, UniDS can alternatively handle chit-chat and task-oriented dialogues in a unified framework. Experimental results demonstrate that the proposed UniDS works comparably well as the pure chit-chat system, and it outperforms state-of-the-art task-oriented dialogue systems. More importantly, UniDS achieves better robustness as it is able to smoothly switch between two types of dialogues. These results demonstrate the feasibility and potential of building an one-for-all dialogue system.

cs.CL↗

arXiv2021

Uncertainty-Aware Balancing for Multilingual and Multi-Domain Neural Machine Translation Training

Learning multilingual and multi-domain translation model is challenging as the heterogeneous and imbalanced data make the model converge inconsistently over different corpora in real world. One common practice is to adjust the share of each corpus in the training, so that the learning process is balanced and low-resource cases can benefit from the high resource ones. However, automatic balancing methods usually depend on the intra- and inter-dataset characteristics, which is usually agnostic or requires human priors. In this work, we propose an approach, MultiUAT, that dynamically adjusts the training data usage based on the model's uncertainty on a small set of trusted clean data for multi-corpus machine translation. We experiments with two classes of uncertainty measures on multilingual (16 languages with 4 settings) and multi-domain settings (4 for in-domain and 2 for out-of-domain on English-German translation) and demonstrate our approach MultiUAT substantially outperforms its baselines, including both static and dynamic strategies. We analyze the cross-domain transfer and show the deficiency of static and similarity based methods.

cs.CL↗

arXiv2021

Acoustic Möbius insulators from projective symmetry

Symmetry plays a critical role in classifying phases of matter. This is exemplified by how crystalline symmetries enrich the topological classification of materials and enable unconventional phenomena in topologically nontrivial ones. After an extensive study over the past decade, the list of topological crystalline insulators and semimetals seems to be exhaustive and concluded. However, in the presence of gauge symmetry, common but not limited to artificial crystals, the algebraic structure of crystalline symmetries needs to be projectively represented, giving rise to unprecedented topological physics. Here we demonstrate this novel idea by exploiting a projective translation symmetry and constructing a variety of Möbius-twisted topological phases. Experimentally, we realize two Möbius insulators in acoustic crystals for the first time: a two-dimensional one of first-order band topology and a three-dimensional one of higher-order band topology. We observe unambiguously the peculiar Möbius edge and hinge states via real-space visualization of their localiztions, momentum-space spectroscopy of their 4π periodicity, and phase-space winding of their projective translation eigenvalues. Not only does our work open a new avenue for artificial systems under the interplay between gauge and crystalline symmetries, but it also initializes a new framework for topological physics from projective symmetry.

cond-mat.mes-hall↗

arXiv2021

Participation Analysis in Impedance Models: The Grey-Box Approach for Power System Stability

This paper develops a grey-box approach to small-signal stability analysis of complex power systems that facilitates root-cause tracing without requiring disclosure of the full details of the internal control structure of apparatus connected to the system. The grey-box enables participation analysis in impedance models, which is popular in power electronics and increasingly accepted in power systems for stability analysis. The Impedance participation factor is proposed and defined in terms of the residue of the whole-system admittance matrix. It is proved that, the so defined impedance participation factor equals the sensitivity of the whole-system eigenvalue with respect to apparatus impedance. The classic state participation factor is related to the impedance participation factor via a chain-rule. Based on the chain-rule, a three-layer grey-box approach, with three degrees of transparency, is proposed for root-cause tracing to different depths, i.e. apparatus, states, and parameters, according to the available information. The association of impedance participation factor with eigenvalue sensitivity points to the re-tuning that would stabilize the system. The impedance participation factor can be measured in the field or calculated from the black-box impedance spectra with little prior knowledge required.

eess.SY↗

arXiv2021

Mapping of Dynamics between Mechanical and Electrical Ports in SG-IBR Composite Grids

Power grids are traditionally dominated by synchronous generators (SGs) but are currently undergoing a major transformation through the increasing integration of inverter-based resources (IBRs). The SG-dominated grid is traditionally analyzed in a mechanical-centric view which ignores fast electrical dynamics and focuses on the torque-speed dynamics. By contrast, analysis of the emergent IBR-dominated grid usually takes the electrical-centric view which focuses on the voltage-current interaction. In this article, a port-mapping method is proposed to fill the gap between these approaches and combine them in a unified model. Specifically, the mechanical dynamics are mapped to the electrical impedance seen at the electrical port; and the electrical dynamics are also mapped to the torque coefficient seen at the mechanical port. The bidirectional mapping gives additional flexibility and insights to analyze the sub-system interactions in whole-system dynamics and guide the tuning of parameter. Application of the proposed method is illustrated in three cases with increasing scales, namely a single-SG-infinite-bus system, a single-IBR-weak-grid system, and a modified IEEE 14-bus SG-IBR composite system.

eess.SY↗

arXiv2021

Automated Generation of Interorganizational Disaster Response Networks through Information Extraction

When a disaster occurs, maintaining and restoring community lifelines subsequently require collective efforts from various stakeholders. Aiming at reducing the efforts associated with generating Stakeholder Collaboration Networks (SCNs), this paper proposes a systematic approach to reliable information extraction for stakeholder collaboration and automated network generation. Specifically, stakeholders and their interactions are extracted from texts through Named Entity Recognition (NER), one of the techniques in natural language processing. Once extracted, the collaboration information is transformed into structured datasets to generate the SCNs automatically. A case study of stakeholder collaboration during Hurricane Harvey was investigated and it had demonstrated the feasibility and applicability of the proposed method. Hence, the proposed approach was proved to significantly reduce practitioners' interpretation and data collection workloads. In the end, discussions and future work are provided.

cs.SI↗