Searcharxiv⌕ Search

arXiv subjects

Nan Yang

Publications and source records attributed to Nan Yang.

At least 109 records · Page 6Linked to original sources
arXiv2023

FIRe: Fast Inverse Rendering using Directional and Signed Distance Functions

Neural 3D implicit representations learn priors that are useful for diverse applications, such as single- or multiple-view 3D reconstruction. A major downside of existing approaches while rendering an image is that they require evaluating the network multiple times per camera ray so that the high computational time forms a bottleneck for downstream applications. We address this problem by introducing a novel neural scene representation that we call the directional distance function (DDF). To this end, we learn a signed distance function (SDF) along with our DDF model to represent a class of shapes. Specifically, our DDF is defined on the unit sphere and predicts the distance to the surface along any given direction. Therefore, our DDF allows rendering images with just a single network evaluation per camera ray. Based on our DDF, we present a novel fast algorithm (FIRe) to reconstruct 3D shapes given a posed depth map. We evaluate our proposed method on 3D reconstruction from single-view depth images, where we empirically show that our algorithm reconstructs 3D shapes more accurately and it is more than 15 times faster (per iteration) than competing methods.

cs.CV↗
arXiv2023

Learning to Rank in Generative Retrieval

Generative retrieval stands out as a promising new paradigm in text retrieval that aims to generate identifier strings of relevant passages as the retrieval target. This generative paradigm taps into powerful generative language models, distinct from traditional sparse or dense retrieval methods. However, only learning to generate is insufficient for generative retrieval. Generative retrieval learns to generate identifiers of relevant passages as an intermediate goal and then converts predicted identifiers into the final passage rank list. The disconnect between the learning objective of autoregressive models and the desired passage ranking target leads to a learning gap. To bridge this gap, we propose a learning-to-rank framework for generative retrieval, dubbed LTRGR. LTRGR enables generative retrieval to learn to rank passages directly, optimizing the autoregressive model toward the final passage ranking target via a rank loss. This framework only requires an additional learning-to-rank training phase to enhance current generative retrieval systems and does not add any burden to the inference stage. We conducted experiments on three public benchmarks, and the results demonstrate that LTRGR achieves state-of-the-art performance among generative retrieval methods. The code and checkpoints are released at https://github.com/liyongqi67/LTRGR.

cs.CL↗
arXiv2023

Event-driven Real-time Retrieval in Web Search

Information retrieval in real-time search presents unique challenges distinct from those encountered in classical web search. These challenges are particularly pronounced due to the rapid change of user search intent, which is influenced by the occurrence and evolution of breaking news events, such as earthquakes, elections, and wars. Previous dense retrieval methods, which primarily focused on static semantic representation, lack the capacity to capture immediate search intent, leading to inferior performance in retrieving the most recent event-related documents in time-sensitive scenarios. To address this issue, this paper expands the query with event information that represents real-time search intent. The Event information is then integrated with the query through a cross-attention mechanism, resulting in a time-context query representation. We further enhance the model's capacity for event representation through multi-task training. Since publicly available datasets such as MS-MARCO do not contain any event information on the query side and have few time-sensitive queries, we design an automatic data collection and annotation pipeline to address this issue, which includes ModelZoo-based Coarse Annotation and LLM-driven Fine Annotation processes. In addition, we share the training tricks such as two-stage training and hard negative sampling. Finally, we conduct a set of offline experiments on a million-scale production dataset to evaluate our approach and deploy an A/B testing in a real online system to verify the performance. Extensive experimental results demonstrate that our proposed approach significantly outperforms existing state-of-the-art baseline methods.

cs.IR↗
arXiv2023

Quantum synchronization via Active-Passive-Decomposition configuration: An open quantum system study

In this paper, we study the synchronization of dissipative quantum harmonic oscillators in the framework of quantum open system via the Active-Passive Decomposition (APD) configuration. We show that two or more quantum systems may be synchronized when the quantum systems of interest are embedded in dissipative environments and influenced by a common classical system. Such a classical system is typically termed as a controller, which (1) can drive quantum systems to cross different regimes (e.g., from periodic to chaotic motions) and (2) constructs the so-called Active-Passive Decomposition configuration such that all the quantum objects under consideration may be synchronized. The main findings of this paper is that we demonstrate that the complete synchronizations measured by the standard quantum deviation may be achieved for both stable regimes (quantum limit circles) and unstable regimes (quantum chaotic motions). As an example, we numerically show in an optomechanical setup that the complete synchronization can be realized in quantum mechanical resonators.

quant-ph↗
arXiv2023

Time-Frequency Localization Characteristics of the Delay-Doppler Plane Orthogonal Pulse

The orthogonal delay-Doppler (DD) division multiplexing (ODDM) modulation has recently been proposed as a promising solution for ensuring reliable communications in high mobility scenarios. In this work, we investigate the time-frequency (TF) localization characteristics of the DD plane orthogonal pulse (DDOP), which is the prototype pulse of ODDM modulation. The TF localization characteristics examine how concentrated or spread out the energy of a pulse is in the joint TF domain. We first derive the TF localization metric, TF area (TFA), for the DDOP. Based on this result, we provide insights into the energy spread of the DDOP in the joint TF domain. Then, we delve into the potential advantages of the DDOP due to its energy spread, particularly in terms of leveraging both time and frequency diversities, and enabling high-resolution sensing. Furthermore, we determine the TFA for the recently proposed generalized design of the DDOP. Finally, we validate our analysis based on numerical results and show that the energy spread for the generalized design of the DDOP in the joint TF domain exhibits a step-wise increase as the duration of sub-pulses increases.

eess.SP↗
arXiv2023

Analysis of Molecule Harvesting by Heterogeneous Receptors on MC Transmitters

This paper designs a molecule harvesting transmitter (TX) model, where the surface of a spherical TX is covered by heterogeneous receptors with different sizes and arbitrary locations. If molecules hit any receptor, they are absorbed by the TX immediately. Within the TX, molecules are stored in vesicles that are continuously generated and released by the TX via the membrane fusion process. Considering a transparent receiver (RX) and molecular degradation during the propagation from the TX to the RX, we derive the molecule release rate and the fraction of molecules absorbed by the TX as well as the received signal at the RX. Notably, this analytical result is applicable for different numbers, sizes, and locations of receptors, and its accuracy is verified via particle-based simulations. Numerical results show that different vesicle generation rates result in the same number of molecules absorbed by the TX, but different peak received signals at the RX.

cs.ET↗
arXiv2023

Fine-Tuning LLaMA for Multi-Stage Text Retrieval

The effectiveness of multi-stage text retrieval has been solidly demonstrated since before the era of pre-trained language models. However, most existing studies utilize models that predate recent advances in large language models (LLMs). This study seeks to explore potential improvements that state-of-the-art LLMs can bring. We conduct a comprehensive study, fine-tuning the latest LLaMA model both as a dense retriever (RepLLaMA) and as a pointwise reranker (RankLLaMA) for both passage retrieval and document retrieval using the MS MARCO datasets. Our findings demonstrate that the effectiveness of large language models indeed surpasses that of smaller models. Additionally, since LLMs can inherently handle longer contexts, they can represent entire documents holistically, obviating the need for traditional segmenting and pooling strategies. Furthermore, evaluations on BEIR demonstrate that our RepLLaMA-RankLLaMA pipeline exhibits strong zero-shot effectiveness. Model checkpoints from this study are available on HuggingFace.

cs.IR↗
arXiv2023

Query2doc: Query Expansion with Large Language Models

This paper introduces a simple yet effective query expansion approach, denoted as query2doc, to improve both sparse and dense retrieval systems. The proposed method first generates pseudo-documents by few-shot prompting large language models (LLMs), and then expands the query with generated pseudo-documents. LLMs are trained on web-scale text corpora and are adept at knowledge memorization. The pseudo-documents from LLMs often contain highly relevant information that can aid in query disambiguation and guide the retrievers. Experimental results demonstrate that query2doc boosts the performance of BM25 by 3% to 15% on ad-hoc IR datasets, such as MS-MARCO and TREC DL, without any model fine-tuning. Furthermore, our method also benefits state-of-the-art dense retrievers in terms of both in-domain and out-of-domain results.

cs.IR↗
arXiv2023

Secure Short-Packet Transmission with Aerial Relaying: Blocklength and Trajectory Co-Design

In this paper, we propose a secure short-packet communication (SPC) system involving an unmanned aerial vehicle (UAV)-aided relay in the presence of a terrestrial passive eavesdropper. The considered system, which is applicable to various next-generation Internet-of-Things (IoT) networks, exploits a UAV as a mobile relay, facilitating the reliable and secure exchange of intermittent short packets between a pair of remote IoT devices with strict latency. Our objective is to improve the overall secrecy throughput performance of the system by carefully designing key parameters such as the coding blocklengths and the UAV trajectory. However, this inherently poses a challenging optimization problem that is difficult to solve optimally. To address the issue, we propose a low-complexity algorithm inspired by the block successive convex approximation approach, where we divide the original problem into two subproblems and solve them alternately until convergence. Numerical results demonstrate that the proposed design achieves significant performance improvements relative to other benchmarks, and offer valuable insights into determining appropriate coding blocklengths and UAV trajectory.

cs.IT↗
arXiv2023

Project Aria: A New Tool for Egocentric Multi-Modal AI Research

Egocentric, multi-modal data as available on future augmented reality (AR) devices provides unique challenges and opportunities for machine perception. These future devices will need to be all-day wearable in a socially acceptable form-factor to support always available, context-aware and personalized AI applications. Our team at Meta Reality Labs Research built the Aria device, an egocentric, multi-modal data recording and streaming device with the goal to foster and accelerate research in this area. In this paper, we describe the Aria device hardware including its sensor configuration and the corresponding software tools that enable recording and processing of such data.

cs.HC↗
arXiv2023

Heterogeneous Receptors - Based Molecule Harvesting in MC: Analysis for ISI Mitigation and Energy Efficiency

This paper investigates a spherical transmitter (TX) with a membrane covered by heterogeneous receptors of varying sizes and arbitrary locations for molecular communication (MC), where molecules are encapsulated within vesicles and released from the TX through membrane fusion. Assuming continuous vesicle generation at the TX and a transparent receiver (RX), we calculate the molecule release rate, the fraction of absorbed molecules at the TX, and the received signal at the RX. All obtained analytical expressions are functions of all receptors locations and sizes, and are validated by particle-based simulations. Our numerical results indicate that evenly distributed receptors on the TX membrane can absorb more molecules than randomly distributed receptors or a single receptor. Furthermore, inspired by the autoreceptor functionality in synaptic communication, we incorporate a negative feedback mechanism (NFM) at the TX, such that molecule release stops after a certain period. We then derive the fraction of molecules that can be reused for the subsequent emissions when considering both NFM and molecule harvesting. Our numerical results demonstrate that incorporating NFM can reduce inter-symbol interference (ISI) while maintaining the same peak received signal as the case without NFM. Additionally, our results show that TXs incorporating both molecule harvesting and NFM can achieve a higher energy efficiency and lower error probability than TXs employing only molecule harvesting or neither functionality.

cs.ET↗
arXiv2023

Age of Information in Downlink Systems: Broadcast or Unicast Transmission?

We analytically decide whether the broadcast transmission scheme or the unicast transmission scheme achieves the optimal age of information (AoI) performance of a multiuser system where a base station (BS) generates and transmits status updates to multiple user equipments (UEs). In the broadcast transmission scheme, the status update for all UEs is jointly encoded into a packet for transmission, while in the unicast transmission scheme, the status update for each UE is encoded individually and transmitted by following the round robin policy. For both transmission schemes, we examine three packet management strategies, namely the non-preemption strategy, the preemption in buffer strategy, and the preemption in serving strategy. We first derive new closed-form expressions for the average AoI achieved by two transmission schemes with three packet management strategies. Based on them, we compare the AoI performance of two transmission schemes in two systems, namely, the remote control system and the dynamic system. Aided by simulation results, we verify our analysis and investigate the impact of system parameters on the average AoI. For example, the unicast transmission scheme is more appropriate for the system with a large number UEs. Otherwise, the broadcast transmission scheme is more appropriate.

cs.IT↗
arXiv2023

Multiview Identifiers Enhanced Generative Retrieval

Instead of simply matching a query to pre-existing passages, generative retrieval generates identifier strings of passages as the retrieval target. At a cost, the identifier must be distinctive enough to represent a passage. Current approaches use either a numeric ID or a text piece (such as a title or substrings) as the identifier. However, these identifiers cannot cover a passage's content well. As such, we are motivated to propose a new type of identifier, synthetic identifiers, that are generated based on the content of a passage and could integrate contextualized information that text pieces lack. Furthermore, we simultaneously consider multiview identifiers, including synthetic identifiers, titles, and substrings. These views of identifiers complement each other and facilitate the holistic ranking of passages from multiple perspectives. We conduct a series of experiments on three public datasets, and the results indicate that our proposed approach performs the best in generative retrieval, demonstrating its effectiveness and robustness.

cs.CL↗
arXiv2023

Incremental Dense Reconstruction from Monocular Video with Guided Sparse Feature Volume Fusion

Incrementally recovering 3D dense structures from monocular videos is of paramount importance since it enables various robotics and AR applications. Feature volumes have recently been shown to enable efficient and accurate incremental dense reconstruction without the need to first estimate depth, but they are not able to achieve as high of a resolution as depth-based methods due to the large memory consumption of high-resolution feature volumes. This letter proposes a real-time feature volume-based dense reconstruction method that predicts TSDF (Truncated Signed Distance Function) values from a novel sparsified deep feature volume, which is able to achieve higher resolutions than previous feature volume-based methods, and is favorable in large-scale outdoor scenarios where the majority of voxels are empty. An uncertainty-aware multi-view stereo (MVS) network is leveraged to infer initial voxel locations of the physical surface in a sparse feature volume. Then for refining the recovered 3D geometry, deep features are attentively aggregated from multiview images at potential surface locations, and temporally fused. Besides achieving higher resolutions than before, our method is shown to produce more complete reconstructions with finer detail in many cases. Extensive evaluations on both public and self-collected datasets demonstrate a very competitive real-time reconstruction result for our method compared to state-of-the-art reconstruction methods in both indoor and outdoor settings.

cs.CV↗
arXiv2023

Combining Adversaries with Anti-adversaries in Training

Adversarial training is an effective learning technique to improve the robustness of deep neural networks. In this study, the influence of adversarial training on deep learning models in terms of fairness, robustness, and generalization is theoretically investigated under more general perturbation scope that different samples can have different perturbation directions (the adversarial and anti-adversarial directions) and varied perturbation bounds. Our theoretical explorations suggest that the combination of adversaries and anti-adversaries (samples with anti-adversarial perturbations) in training can be more effective in achieving better fairness between classes and a better tradeoff between robustness and generalization in some typical learning scenarios (e.g., noisy label learning and imbalance learning) compared with standard adversarial training. On the basis of our theoretical findings, a more general learning objective that combines adversaries and anti-adversaries with varied bounds on each training sample is presented. Meta learning is utilized to optimize the combination weights. Experiments on benchmark datasets under different learning scenarios verify our theoretical findings and the effectiveness of the proposed methodology.

cs.LG↗
arXiv2023

SimLM: Pre-training with Representation Bottleneck for Dense Passage Retrieval

In this paper, we propose SimLM (Similarity matching with Language Model pre-training), a simple yet effective pre-training method for dense passage retrieval. It employs a simple bottleneck architecture that learns to compress the passage information into a dense vector through self-supervised pre-training. We use a replaced language modeling objective, which is inspired by ELECTRA, to improve the sample efficiency and reduce the mismatch of the input distribution between pre-training and fine-tuning. SimLM only requires access to unlabeled corpus, and is more broadly applicable when there are no labeled data or queries. We conduct experiments on several large-scale passage retrieval datasets, and show substantial improvements over strong baselines under various settings. Remarkably, SimLM even outperforms multi-vector approaches such as ColBERTv2 which incurs significantly more storage cost. Our code and model check points are available at https://github.com/microsoft/unilm/tree/master/simlm .

cs.IR↗
arXiv2023

Behind the Scenes: Density Fields for Single View Reconstruction

Inferring a meaningful geometric scene representation from a single image is a fundamental problem in computer vision. Approaches based on traditional depth map prediction can only reason about areas that are visible in the image. Currently, neural radiance fields (NeRFs) can capture true 3D including color, but are too complex to be generated from a single image. As an alternative, we propose to predict implicit density fields. A density field maps every location in the frustum of the input image to volumetric density. By directly sampling color from the available views instead of storing color in the density field, our scene representation becomes significantly less complex compared to NeRFs, and a neural network can predict it in a single forward pass. The prediction network is trained through self-supervision from only video data. Our formulation allows volume rendering to perform both depth prediction and novel view synthesis. Through experiments, we show that our method is able to predict meaningful geometry for regions that are occluded in the input image. Additionally, we demonstrate the potential of our approach on three datasets for depth prediction and novel-view synthesis.

cs.CV↗
arXiv2023

Inference with Reference: Lossless Acceleration of Large Language Models

We propose LLMA, an LLM accelerator to losslessly speed up Large Language Model (LLM) inference with references. LLMA is motivated by the observation that there are abundant identical text spans between the decoding result by an LLM and the reference that is available in many real world scenarios (e.g., retrieved documents). LLMA first selects a text span from the reference and copies its tokens to the decoder and then efficiently checks the tokens' appropriateness as the decoding result in parallel within one decoding step. The improved computational parallelism allows LLMA to achieve over 2x speed-up for LLMs with identical generation results as greedy decoding in many practical generation scenarios where significant overlap between in-context reference and outputs exists (e.g., search engines and multi-turn conversations).

cs.CL↗