arXiv subjects

Tian Lan

Publications and source records attributed to Tian Lan.

At least 199 records · Page 11Linked to original sources

arXiv2023

ReMIX: Regret Minimization for Monotonic Value Function Factorization in Multiagent Reinforcement Learning

Value function factorization methods have become a dominant approach for cooperative multiagent reinforcement learning under a centralized training and decentralized execution paradigm. By factorizing the optimal joint action-value function using a monotonic mixing function of agents' utilities, these algorithms ensure the consistency between joint and local action selections for decentralized decision-making. Nevertheless, the use of monotonic mixing functions also induces representational limitations. Finding the optimal projection of an unrestricted mixing function onto monotonic function classes is still an open problem. To this end, we propose ReMIX, formulating this optimal projection problem for value function factorization as a regret minimization over the projection weights of different state-action values. Such an optimization problem can be relaxed and solved using the Lagrangian multiplier method to obtain the close-form optimal projection weights. By minimizing the resulting policy regret, we can narrow the gap between the optimal and the restricted monotonic mixing functions, thus obtaining an improved monotonic value function factorization. Our experimental results on Predator-Prey and StarCraft Multiagent Challenge environments demonstrate the effectiveness of our method, indicating the better capabilities of handling environments with non-monotonic value functions.

cs.LG↗

arXiv2023

Scalability Bottlenecks in Multi-Agent Reinforcement Learning Systems

Multi-Agent Reinforcement Learning (MARL) is a promising area of research that can model and control multiple, autonomous decision-making agents. During online training, MARL algorithms involve performance-intensive computations such as exploration and exploitation phases originating from large observation-action space belonging to multiple agents. In this article, we seek to characterize the scalability bottlenecks in several popular classes of MARL algorithms during their training phases. Our experimental results reveal new insights into the key modules of MARL algorithms that limit the scalability, and outline potential strategies that may help address these performance issues.

cs.MA↗

arXiv2023

PAC: Assisted Value Factorisation with Counterfactual Predictions in Multi-Agent Reinforcement Learning

Multi-agent reinforcement learning (MARL) has witnessed significant progress with the development of value function factorization methods. It allows optimizing a joint action-value function through the maximization of factorized per-agent utilities due to monotonicity. In this paper, we show that in partially observable MARL problems, an agent's ordering over its own actions could impose concurrent constraints (across different states) on the representable function class, causing significant estimation error during training. We tackle this limitation and propose PAC, a new framework leveraging Assistive information generated from Counterfactual Predictions of optimal joint action selection, which enable explicit assistance to value function factorization through a novel counterfactual loss. A variational inference-based information encoding method is developed to collect and encode the counterfactual predictions from an estimated baseline. To enable decentralized execution, we also derive factorized per-agent policies inspired by a maximum-entropy MARL framework. We evaluate the proposed PAC on multi-agent predator-prey and a set of StarCraft II micromanagement tasks. Empirical results demonstrate improved results of PAC over state-of-the-art value-based and policy-based multi-agent reinforcement learning algorithms on all benchmarks.

cs.MA↗

arXiv2023

TWR-MCAE: A Data Augmentation Method for Through-the-Wall Radar Human Motion Recognition

To solve the problems of reduced accuracy and prolonging convergence time of through-the-wall radar (TWR) human motion due to wall attenuation, multipath effect, and system interference, we propose a multilink auto-encoding neural network (TWR-MCAE) data augmentation method. Specifically, the TWR-MCAE algorithm is jointly constructed by a singular value decomposition (SVD)-based data preprocessing module, an improved coordinate attention module, a compressed sensing learnable iterative shrinkage threshold reconstruction algorithm (LISTA) module, and an adaptive weight module. The data preprocessing module achieves wall clutter, human motion features, and noise subspaces separation. The improved coordinate attention module achieves clutter and noise suppression. The LISTA module achieves human motion feature enhancement. The adaptive weight module learns the weights and fuses the three subspaces. The TWR-MCAE can suppress the low-rank characteristics of wall clutter and enhance the sparsity characteristics in human motion at the same time. It can be linked before the classification step to improve the feature extraction capability without adding other prior knowledge or recollecting more data. Experiments show that the proposed algorithm gets a better peak signal-to-noise ratio (PSNR), which increases the recognition accuracy and speeds up the training process of the back-end classifiers.

eess.SP↗

arXiv2022

Momentum Decoding: Open-ended Text Generation As Graph Exploration

Open-ended text generation with autoregressive language models (LMs) is one of the core tasks in natural language processing. However, maximization-based decoding methods (e.g., greedy/beam search) often lead to the degeneration problem, i.e., the generated text is unnatural and contains undesirable repetitions. Existing solutions to this problem either introduce randomness prone to incoherence or require a look-ahead mechanism that demands extra computational overhead. In this study, we formulate open-ended text generation from a new perspective, i.e., we view it as an exploration process within a directed graph. Thereby, we understand the phenomenon of degeneration as circular loops within the directed graph. Based on our formulation, we propose a novel decoding method -- \textit{momentum decoding} -- which encourages the LM to \textit{greedily} explore new nodes outside the current graph. Meanwhile, it also allows the LM to return to the existing nodes with a momentum downgraded by a pre-defined resistance function. We extensively test our approach on three benchmarks from different domains through automatic and human evaluations. The results show that momentum decoding performs comparably with the current state of the art while enjoying notably improved inference speed and computation FLOPs. Furthermore, we conduct a detailed analysis to reveal the merits and inner workings of our approach. Our codes and other related resources are publicly available at https://github.com/gmftbyGMFTBY/MomentumDecoding.

cs.CL↗

arXiv2022

Constructing the LG/CY isomorphism between $tt^*$ geometries

For a nondegenerate homogeneous polynomial $f\in\mathbb{C}[z_0, \dots, z_{n+1}]$ with degree $n+2$, we can obtain a $tt^*$ structure from the Landau-Ginzburg model $(\C^{n+2}, f)$ and a (new) $tt^*$ structure on the Calabi-Yau hypersurface defined by the zero locus of $f$ in $\C P^{n+1}$. We can prove that the big residue map considered by Steenbrink gives an isomorphism between the two $tt^*$ structures. We also build the correspondence for non-Calabi-Yau cases, and it turns out that only partial structure can be preserved. As an application, we show that the $tt^*$ geometry structure of Landau-Ginzburg model on relavant deformation space uniquely determines the $tt^*$ geometry structure on Calabi-Yau side. This explains the folklore conclusion in physical literature. This result is based on our early work \cite{FLY}.

math.AG↗

arXiv2022

A Contrastive Framework for Neural Text Generation

Text generation is of great importance to many natural language processing applications. However, maximization-based decoding methods (e.g. beam search) of neural language models often lead to degenerate solutions -- the generated text is unnatural and contains undesirable repetitions. Existing approaches introduce stochasticity via sampling or modify training objectives to decrease probabilities of certain tokens (e.g., unlikelihood training). However, they often lead to solutions that lack coherence. In this work, we show that an underlying reason for model degeneration is the anisotropic distribution of token representations. We present a contrastive solution: (i) SimCTG, a contrastive training objective to calibrate the model's representation space, and (ii) a decoding method -- contrastive search -- to encourage diversity while maintaining coherence in the generated text. Extensive experiments and analyses on three benchmarks from two languages demonstrate that our proposed approach significantly outperforms current state-of-the-art text generation methods as evaluated by both human and automatic metrics.

cs.CL↗

arXiv2022

Language Models Can See: Plugging Visual Controls in Text Generation

Generative language models (LMs) such as GPT-2/3 can be prompted to generate text with remarkable quality. While they are designed for text-prompted generation, it remains an open question how the generation process could be guided by modalities beyond text such as images. In this work, we propose a training-free framework, called MAGIC (iMAge-Guided text generatIon with CLIP), for plugging in visual controls in the generation process and enabling LMs to perform multimodal tasks (e.g., image captioning) in a zero-shot manner. MAGIC is a simple yet efficient plug-and-play framework, which directly combines an off-the-shelf LM (i.e., GPT-2) and an image-text matching model (i.e., CLIP) for image-grounded text generation. During decoding, MAGIC influences the generation of the LM by introducing a CLIP-induced score, called magic score, which regularizes the generated result to be semantically related to a given image while being coherent to the previously generated context. Notably, the proposed decoding scheme does not involve any gradient update operation, therefore being computationally efficient. On the challenging task of zero-shot image captioning, MAGIC outperforms the state-of-the-art method by notable margins with a nearly 27 times decoding speedup. MAGIC is a flexible framework and is theoretically compatible with any text generation tasks that incorporate image grounding. In the experiments, we showcase that it is also capable of performing visually grounded story generation given both an image and a text prompt.

cs.CV↗

arXiv2022

TaCL: Improving BERT Pre-training with Token-aware Contrastive Learning

Masked language models (MLMs) such as BERT and RoBERTa have revolutionized the field of Natural Language Understanding in the past few years. However, existing pre-trained MLMs often output an anisotropic distribution of token representations that occupies a narrow subset of the entire representation space. Such token representations are not ideal, especially for tasks that demand discriminative semantic meanings of distinct tokens. In this work, we propose TaCL (Token-aware Contrastive Learning), a novel continual pre-training approach that encourages BERT to learn an isotropic and discriminative distribution of token representations. TaCL is fully unsupervised and requires no additional data. We extensively test our approach on a wide range of English and Chinese benchmarks. The results show that TaCL brings consistent and notable improvements over the original BERT model. Furthermore, we conduct detailed analysis to reveal the merits and inner-workings of our approach.

cs.CL↗

arXiv2022

Exploring Dense Retrieval for Dialogue Response Selection

Recent progress in deep learning has continuously improved the accuracy of dialogue response selection. In particular, sophisticated neural network architectures are leveraged to capture the rich interactions between dialogue context and response candidates. While remarkably effective, these models also bring in a steep increase in computational cost. Consequently, such models can only be used as a re-rank module in practice. In this study, we present a solution to directly select proper responses from a large corpus or even a nonparallel corpus that only consists of unpaired sentences, using a dense retrieval model. To push the limits of dense retrieval, we design an interaction layer upon the dense retrieval models and apply a set of tailor-designed learning strategies. Our model shows superiority over strong baselines on the conventional re-rank evaluation setting, which is remarkable given its efficiency. To verify the effectiveness of our approach in realistic scenarios, we also conduct full-rank evaluation, where the target is to select proper responses from a full candidate pool that may contain millions of candidates and evaluate them fairly through human annotations. Our proposed model notably outperforms pipeline baselines that integrate fast recall and expressive re-rank modules. Human evaluation results show that enlarging the candidate pool with nonparallel corpora improves response quality further.

cs.CL↗

arXiv2022

Cross-Lingual Phrase Retrieval

Cross-lingual retrieval aims to retrieve relevant text across languages. Current methods typically achieve cross-lingual retrieval by learning language-agnostic text representations in word or sentence level. However, how to learn phrase representations for cross-lingual phrase retrieval is still an open problem. In this paper, we propose XPR, a cross-lingual phrase retriever that extracts phrase representations from unlabeled example sentences. Moreover, we create a large-scale cross-lingual phrase retrieval dataset, which contains 65K bilingual phrase pairs and 4.2M example sentences in 8 English-centric language pairs. Experimental results show that XPR outperforms state-of-the-art baselines which utilize word-level or sentence-level representations. XPR also shows impressive zero-shot transferability that enables the model to perform retrieval in an unseen language pair during training. Our dataset, code, and trained models are publicly available at www.github.com/cwszz/XPR/.

cs.CL↗

arXiv2022

SAFARI: Sparsity enabled Federated Learning with Limited and Unreliable Communications

Federated learning (FL) enables edge devices to collaboratively learn a model in a distributed fashion. Many existing researches have focused on improving communication efficiency of high-dimensional models and addressing bias caused by local updates. However, most of FL algorithms are either based on reliable communications or assume fixed and known unreliability characteristics. In practice, networks could suffer from dynamic channel conditions and non-deterministic disruptions, with time-varying and unknown characteristics. To this end, in this paper we propose a sparsity enabled FL framework with both communication efficiency and bias reduction, termed as SAFARI. It makes novel use of a similarity among client models to rectify and compensate for bias that is resulted from unreliable communications. More precisely, sparse learning is implemented on local clients to mitigate communication overhead, while to cope with unreliable communications, a similarity-based compensation method is proposed to provide surrogates for missing model updates. We analyze SAFARI under bounded dissimilarity and with respect to sparse models. It is demonstrated that SAFARI under unreliable communications is guaranteed to converge at the same rate as the standard FedAvg with perfect communications. Implementations and evaluations on CIFAR-10 dataset validate the effectiveness of SAFARI by showing that it can achieve the same convergence speed and accuracy as FedAvg with perfect communications, with up to 80% of the model weights being pruned and a high percentage of client updates missing in each round.

cs.DC↗

arXiv2022

Efficient Video Instance Segmentation via Tracklet Query and Proposal

Video Instance Segmentation (VIS) aims to simultaneously classify, segment, and track multiple object instances in videos. Recent clip-level VIS takes a short video clip as input each time showing stronger performance than frame-level VIS (tracking-by-segmentation), as more temporal context from multiple frames is utilized. Yet, most clip-level methods are neither end-to-end learnable nor real-time. These limitations are addressed by the recent VIS transformer (VisTR) which performs VIS end-to-end within a clip. However, VisTR suffers from long training time due to its frame-wise dense attention. In addition, VisTR is not fully end-to-end learnable in multiple video clips as it requires a hand-crafted data association to link instance tracklets between successive clips. This paper proposes EfficientVIS, a fully end-to-end framework with efficient training and inference. At the core are tracklet query and tracklet proposal that associate and segment regions-of-interest (RoIs) across space and time by an iterative query-video interaction. We further propose a correspondence learning that makes tracklets linking between clips end-to-end learnable. Compared to VisTR, EfficientVIS requires 15x fewer training epochs while achieving state-of-the-art accuracy on the YouTube-VIS benchmark. Meanwhile, our method enables whole video instance segmentation in a single end-to-end pass without data association at all.

cs.CV↗

arXiv2022

On the Convergence of Heterogeneous Federated Learning with Arbitrary Adaptive Online Model Pruning

One of the biggest challenges in Federated Learning (FL) is that client devices often have drastically different computation and communication resources for local updates. To this end, recent research efforts have focused on training heterogeneous local models obtained by pruning a shared global model. Despite empirical success, theoretical guarantees on convergence remain an open question. In this paper, we present a unifying framework for heterogeneous FL algorithms with {\em arbitrary} adaptive online model pruning and provide a general convergence analysis. In particular, we prove that under certain sufficient conditions and on both IID and non-IID data, these algorithms converges to a stationary point of standard FL for general smooth cost functions, with a convergence rate of $O(\frac{1}{\sqrt{Q}})$. Moreover, we illuminate two key factors impacting convergence: pruning-induced noise and minimum coverage index, advocating a joint design of local pruning masks for efficient training.

cs.LG↗

arXiv2022

A Framework for Server Authentication using Communication Protocol Dialects

In today's world, computer networks have become vulnerable to numerous attacks. In both wireless and wired networks, one of the most common attacks is man-in-the-middle attacks, within which session hijacking, context confusion attacks have been the most attempted. A potential attacker may have enough time to launch an attack targeting these vulnerabilities (such as rerouting the target request to a malicious server or hijacking the traffic). A viable strategy to solve this problem is, by dynamically changing the system properties, configurations and create unique fingerprints to identify the source. However, the existing work of fingerprinting mainly focuses on lower-level properties (e.g IP address), and only these types of properties are restricted for mutation. We develop a novel system, called Verify-Pro, to provide server authentication using communication protocol dialects, that uses a client-server architecture based on network protocols for customizing the communication transactions. For each session, a particular sequence of handshakes will be used as dialects. So, given the context, with the establishment of a one-time username and password, we use the dialects as an authentication mechanism for each request (e.g get filename in FTP) throughout the session, which enforces continuous authentication. Specifically, we leverage a machine learning approach on both client and server machines to trigger a specific dialect that dynamically changes for each request. We implement a prototype of Verify-Pro and evaluate its practicality on standard communication protocols FTP, HTTP & internet of things protocol MQTT. Our experimental results show that by sending misleading information through message packets from an attacker at the application layer, it is possible for the recipient to identify if the sender is genuine or a spoofed one, with a negligible overhead of 0.536%.

cs.CR↗

arXiv2021

A lattice realization of general three-dimensional topological order

Topological orders are a class of phases of matter that beyond the Landau symmetry breaking paradigm. The two (spatial) dimensional (2d) topological orders have been thoroughly studied. It is known that they can be fully classified by a unitary modular tensor category (UMTC) and a chiral central charge c. And a class of 2d topological orders whose boundary are gappable can be systematically constructed by Levin-Wen model whose ground states are string-net condensed states. Previously, the three spatial dimensional topological orders have been classified based on their canonical boundary described by some special unitary fusion 2-category, $2\mathcal{V}ec_G^ω$ or an EF 2-category. However, a lattice realization of a three spatial dimensional topological orders with both canonical boundary and arbitrary boundaries are still lacking. In this paper, we construct a 3d membrane-net model based on spherical fusion 2-category, which can be used to systematically study all general 3d topological order with gapped boundary. The partition function and lattice Hamiltonian of the membrane-net model is constructed based on state sum of spherical fusion 2-category. We also construct the 3d tube algebra of the membrane-net model to study excitations in the model. We conjecture all intrinsic excitations in membrane-net model have a one-to-one correspondence with the irreducible central idempotents (ICI) of the 3d tube algebra. We also provide a universal framework to study mutual statistics of all excitations in 3d topological order through 3d tube algebra. Our approach can be straightforwardly generalized to arbitrary dimension.

cond-mat.str-el↗

arXiv2021

PT-VTON: an Image-Based Virtual Try-On Network with Progressive Pose Attention Transfer

The virtual try-on system has gained great attention due to its potential to give customers a realistic, personalized product presentation in virtualized settings. In this paper, we present PT-VTON, a novel pose-transfer-based framework for cloth transfer that enables virtual try-on with arbitrary poses. PT-VTON can be applied to the fashion industry within minimal modification of existing systems while satisfying the overall visual fashionability and detailed fabric appearance requirements. It enables efficient clothes transferring between model and user images with arbitrary pose and body shape. We implement a prototype of PT-VTON and demonstrate that our system can match or surpass many other approaches when facing a drastic variation of poses by preserving detailed human and fabric characteristic appearances. PT-VTON is shown to outperform alternative approaches both on machine-based quantitative metrics and qualitative results.

cs.CV↗

arXiv2021

Foster Strengths and Circumvent Weaknesses: a Speech Enhancement Framework with Two-branch Collaborative Learning

Recent single-channel speech enhancement methods usually convert waveform to the time-frequency domain and use magnitude/complex spectrum as the optimizing target. However, both magnitude-spectrum-based methods and complex-spectrum-based methods have their respective pros and cons. In this paper, we propose a unified two-branch framework to foster strengths and circumvent weaknesses of different paradigms. The proposed framework could take full advantage of the apparent spectral regularity in magnitude spectrogram and break the bottleneck that magnitude-based methods have suffered. Within each branch, we use collaborative expert block and its variants as substitutes for regular convolution layers. Experiments on TIMIT benchmark demonstrate that our method is superior to existing state-of-the-art ones.

cs.SD↗