Searcharxiv⌕ Search

arXiv subjects

Minsu Kim

Publications and source records attributed to Minsu Kim.

At least 127 records · Page 7Linked to original sources
arXiv2023

DF-3DFace: One-to-Many Speech Synchronized 3D Face Animation with Diffusion

Speech-driven 3D facial animation has gained significant attention for its ability to create realistic and expressive facial animations in 3D space based on speech. Learning-based methods have shown promising progress in achieving accurate facial motion synchronized with speech. However, one-to-many nature of speech-to-3D facial synthesis has not been fully explored: while the lip accurately synchronizes with the speech content, other facial attributes beyond speech-related motions are variable with respect to the speech. To account for the potential variance in the facial attributes within a single speech, we propose DF-3DFace, a diffusion-driven speech-to-3D face mesh synthesis. DF-3DFace captures the complex one-to-many relationships between speech and 3D face based on diffusion. It concurrently achieves aligned lip motion by exploiting audio-mesh synchronization and masked conditioning. Furthermore, the proposed method jointly models identity and pose in addition to facial motions so that it can generate 3D face animation without requiring a reference identity mesh and produce natural head poses. We contribute a new large-scale 3D facial mesh dataset, 3D-HDTF to enable the synthesis of variations in identities, poses, and facial motions of 3D face mesh. Extensive experiments demonstrate that our method successfully generates highly variable facial shapes and motions from speech and simultaneously achieves more realistic facial animation than the state-of-the-art methods.

cs.CV↗
arXiv2023

Green, Quantized Federated Learning over Wireless Networks: An Energy-Efficient Design

In this paper, a green-quantized FL framework, which represents data with a finite precision level in both local training and uplink transmission, is proposed. Here, the finite precision level is captured through the use of quantized neural networks (QNNs) that quantize weights and activations in fixed-precision format. In the considered FL model, each device trains its QNN and transmits a quantized training result to the base station. Energy models for the local training and the transmission with quantization are rigorously derived. To minimize the energy consumption and the number of communication rounds simultaneously, a multi-objective optimization problem is formulated with respect to the number of local iterations, the number of selected devices, and the precision levels for both local training and transmission while ensuring convergence under a target accuracy constraint. To solve this problem, the convergence rate of the proposed FL system is analytically derived with respect to the system control variables. Then, the Pareto boundary of the problem is characterized to provide efficient solutions using the normal boundary inspection method. Design insights on balancing the tradeoff between the two objectives while achieving a target accuracy are drawn from using the Nash bargaining solution and analyzing the derived convergence rate. Simulation results show that the proposed FL framework can reduce energy consumption until convergence by up to 70\% compared to a baseline FL algorithm that represents data with full precision without damaging the convergence rate.

cs.LG↗
arXiv2023

DevFormer: A Symmetric Transformer for Context-Aware Device Placement

In this paper, we present DevFormer, a novel transformer-based architecture for addressing the complex and computationally demanding problem of hardware design optimization. Despite the demonstrated efficacy of transformers in domains including natural language processing and computer vision, their use in hardware design has been limited by the scarcity of offline data. Our approach addresses this limitation by introducing strong inductive biases such as relative positional embeddings and action-permutation symmetricity that effectively capture the hardware context and enable efficient design optimization with limited offline data. We apply DevFoemer to the problem of decoupling capacitor placement and show that it outperforms state-of-the-art methods in both simulated and real hardware, leading to improved performances while reducing the number of components by more than $30\%$. Finally, we show that our approach achieves promising results in other offline contextual learning-based combinatorial optimization tasks.

cs.LG↗
arXiv2023

Meta-SAGE: Scale Meta-Learning Scheduled Adaptation with Guided Exploration for Mitigating Scale Shift on Combinatorial Optimization

This paper proposes Meta-SAGE, a novel approach for improving the scalability of deep reinforcement learning models for combinatorial optimization (CO) tasks. Our method adapts pre-trained models to larger-scale problems in test time by suggesting two components: a scale meta-learner (SML) and scheduled adaptation with guided exploration (SAGE). First, SML transforms the context embedding for subsequent adaptation of SAGE based on scale information. Then, SAGE adjusts the model parameters dedicated to the context embedding for a specific instance. SAGE introduces locality bias, which encourages selecting nearby locations to determine the next location. The locality bias gradually decays as the model is adapted to the target instance. Results show that Meta-SAGE outperforms previous adaptation methods and significantly improves scalability in representative CO tasks. Our source code is available at https://github.com/kaist-silab/meta-sage

cs.LG↗
arXiv2023

Intelligible Lip-to-Speech Synthesis with Speech Units

In this paper, we propose a novel Lip-to-Speech synthesis (L2S) framework, for synthesizing intelligible speech from a silent lip movement video. Specifically, to complement the insufficient supervisory signal of the previous L2S model, we propose to use quantized self-supervised speech representations, named speech units, as an additional prediction target for the L2S model. Therefore, the proposed L2S model is trained to generate multiple targets, mel-spectrogram and speech units. As the speech units are discrete while mel-spectrogram is continuous, the proposed multi-target L2S model can be trained with strong content supervision, without using text-labeled data. Moreover, to accurately convert the synthesized mel-spectrogram into a waveform, we introduce a multi-input vocoder that can generate a clear waveform even from blurry and noisy mel-spectrogram by referring to the speech units. Extensive experimental results confirm the effectiveness of the proposed method in L2S.

cs.SD↗
arXiv2023

Multi-Temporal Lip-Audio Memory for Visual Speech Recognition

Visual Speech Recognition (VSR) is a task to predict a sentence or word from lip movements. Some works have been recently presented which use audio signals to supplement visual information. However, existing methods utilize only limited information such as phoneme-level features and soft labels of Automatic Speech Recognition (ASR) networks. In this paper, we present a Multi-Temporal Lip-Audio Memory (MTLAM) that makes the best use of audio signals to complement insufficient information of lip movements. The proposed method is mainly composed of two parts: 1) MTLAM saves multi-temporal audio features produced from short- and long-term audio signals, and the MTLAM memorizes a visual-to-audio mapping to load stored multi-temporal audio features from visual features at the inference phase. 2) We design an audio temporal model to produce multi-temporal audio features capturing the context of neighboring words. In addition, to construct effective visual-to-audio mapping, the audio temporal models can generate audio features time-aligned with visual features. Through extensive experiments, we validate the effectiveness of the MTLAM achieving state-of-the-art performances on two public VSR datasets.

cs.CV↗
arXiv2023

PartMix: Regularization Strategy to Learn Part Discovery for Visible-Infrared Person Re-identification

Modern data augmentation using a mixture-based technique can regularize the models from overfitting to the training data in various computer vision applications, but a proper data augmentation technique tailored for the part-based Visible-Infrared person Re-IDentification (VI-ReID) models remains unexplored. In this paper, we present a novel data augmentation technique, dubbed PartMix, that synthesizes the augmented samples by mixing the part descriptors across the modalities to improve the performance of part-based VI-ReID models. Especially, we synthesize the positive and negative samples within the same and across different identities and regularize the backbone model through contrastive learning. In addition, we also present an entropy-based mining strategy to weaken the adverse impact of unreliable positive and negative samples. When incorporated into existing part-based VI-ReID model, PartMix consistently boosts the performance. We conduct experiments to demonstrate the effectiveness of our PartMix over the existing VI-ReID methods and provide ablation studies.

cs.CV↗
arXiv2023

Watch or Listen: Robust Audio-Visual Speech Recognition with Visual Corruption Modeling and Reliability Scoring

This paper deals with Audio-Visual Speech Recognition (AVSR) under multimodal input corruption situations where audio inputs and visual inputs are both corrupted, which is not well addressed in previous research directions. Previous studies have focused on how to complement the corrupted audio inputs with the clean visual inputs with the assumption of the availability of clean visual inputs. However, in real life, clean visual inputs are not always accessible and can even be corrupted by occluded lip regions or noises. Thus, we firstly analyze that the previous AVSR models are not indeed robust to the corruption of multimodal input streams, the audio and the visual inputs, compared to uni-modal models. Then, we design multimodal input corruption modeling to develop robust AVSR models. Lastly, we propose a novel AVSR framework, namely Audio-Visual Reliability Scoring module (AV-RelScore), that is robust to the corrupted multimodal inputs. The AV-RelScore can determine which input modal stream is reliable or not for the prediction and also can exploit the more reliable streams in prediction. The effectiveness of the proposed method is evaluated with comprehensive experiments on popular benchmark databases, LRS2 and LRS3. We also show that the reliability scores obtained by AV-RelScore well reflect the degree of corruption and make the proposed model focus on the reliable multimodal representations.

cs.MM↗
arXiv2023

Deep Visual Forced Alignment: Learning to Align Transcription with Talking Face Video

Forced alignment refers to a technology that time-aligns a given transcription with a corresponding speech. However, as the forced alignment technologies have developed using speech audio, they might fail in alignment when the input speech audio is noise-corrupted or is not accessible. We focus on that there is another component that the speech can be inferred from, the speech video (i.e., talking face video). Since the drawbacks of audio-based forced alignment can be complemented using the visual information when the audio signal is under poor condition, we try to develop a novel video-based forced alignment method. However, different from audio forced alignment, it is challenging to develop a reliable visual forced alignment technology for the following two reasons: 1) Visual Speech Recognition (VSR) has a much lower performance compared to audio-based Automatic Speech Recognition (ASR), and 2) the translation from text to video is not reliable, so the method typically used for building audio forced alignment cannot be utilized in developing visual forced alignment. In order to alleviate these challenges, in this paper, we propose a new method that is appropriate for visual forced alignment, namely Deep Visual Forced Alignment (DVFA). The proposed DVFA can align the input transcription (i.e., sentence) with the talking face video without accessing the speech audio. Moreover, by augmenting the alignment task with anomaly case detection, DVFA can detect mismatches between the input transcription and the input video while performing the alignment. Therefore, we can robustly align the text with the talking face video even if there exist error words in the text. Through extensive experiments, we show the effectiveness of the proposed DVFA not only in the alignment task but also in interpreting the outputs of VSR models.

cs.CV↗
arXiv2023

Lip-to-Speech Synthesis in the Wild with Multi-task Learning

Recent studies have shown impressive performance in Lip-to-speech synthesis that aims to reconstruct speech from visual information alone. However, they have been suffering from synthesizing accurate speech in the wild, due to insufficient supervision for guiding the model to infer the correct content. Distinct from the previous methods, in this paper, we develop a powerful Lip2Speech method that can reconstruct speech with correct contents from the input lip movements, even in a wild environment. To this end, we design multi-task learning that guides the model using multimodal supervision, i.e., text and audio, to complement the insufficient word representations of acoustic feature reconstruction loss. Thus, the proposed framework brings the advantage of synthesizing speech containing the right content of multiple speakers with unconstrained sentences. We verify the effectiveness of the proposed method using LRS2, LRS3, and LRW datasets.

cs.SD↗
arXiv2023

On the Tradeoff between Energy, Precision, and Accuracy in Federated Quantized Neural Networks

Deploying federated learning (FL) over wireless networks with resource-constrained devices requires balancing between accuracy, energy efficiency, and precision. Prior art on FL often requires devices to train deep neural networks (DNNs) using a 32-bit precision level for data representation to improve accuracy. However, such algorithms are impractical for resource-constrained devices since DNNs could require execution of millions of operations. Thus, training DNNs with a high precision level incurs a high energy cost for FL. In this paper, a quantized FL framework, that represents data with a finite level of precision in both local training and uplink transmission, is proposed. Here, the finite level of precision is captured through the use of quantized neural networks (QNNs) that quantize weights and activations in fixed-precision format. In the considered FL model, each device trains its QNN and transmits a quantized training result to the base station. Energy models for the local training and the transmission with the quantization are rigorously derived. An energy minimization problem is formulated with respect to the level of precision while ensuring convergence. To solve the problem, we first analytically derive the FL convergence rate and use a line search method. Simulation results show that our FL framework can reduce energy consumption by up to 53% compared to a standard FL model. The results also shed light on the tradeoff between precision, energy, and accuracy in FL over wireless networks.

cs.LG↗
arXiv2023

Sym-NCO: Leveraging Symmetricity for Neural Combinatorial Optimization

Deep reinforcement learning (DRL)-based combinatorial optimization (CO) methods (i.e., DRL-NCO) have shown significant merit over the conventional CO solvers as DRL-NCO is capable of learning CO solvers less relying on problem-specific expert domain knowledge (heuristic method) and supervised labeled data (supervised learning method). This paper presents a novel training scheme, Sym-NCO, which is a regularizer-based training scheme that leverages universal symmetricities in various CO problems and solutions. Leveraging symmetricities such as rotational and reflectional invariance can greatly improve the generalization capability of DRL-NCO because it allows the learned solver to exploit the commonly shared symmetricities in the same CO problem class. Our experimental results verify that our Sym-NCO greatly improves the performance of DRL-NCO methods in four CO tasks, including the traveling salesman problem (TSP), capacitated vehicle routing problem (CVRP), prize collecting TSP (PCTSP), and orienteering problem (OP), without utilizing problem-specific expert domain knowledge. Remarkably, Sym-NCO outperformed not only the existing DRL-NCO methods but also a competitive conventional solver, the iterative local search (ILS), in PCTSP at 240 faster speed. Our source code is available at https://github.com/alstn12088/Sym-NCO.

cs.LG↗
arXiv2022

A Bargaining Game for Personalized, Energy Efficient Split Learning over Wireless Networks

Split learning (SL) is an emergent distributed learning framework which can mitigate the computation and wireless communication overhead of federated learning. It splits a machine learning model into a device-side model and a server-side model at a cut layer. Devices only train their allocated model and transmit the activations of the cut layer to the server. However, SL can lead to data leakage as the server can reconstruct the input data using the correlation between the input and intermediate activations. Although allocating more layers to a device-side model can reduce the possibility of data leakage, this will lead to more energy consumption for resource-constrained devices and more training time for the server. Moreover, non-iid datasets across devices will reduce the convergence rate leading to increased training time. In this paper, a new personalized SL framework is proposed. For this framework, a novel approach for choosing the cut layer that can optimize the tradeoff between the energy consumption for computation and wireless transmission, training time, and data privacy is developed. In the considered framework, each device personalizes its device-side model to mitigate non-iid datasets while sharing the same server-side model for generalization. To balance the energy consumption for computation and wireless transmission, training time, and data privacy, a multiplayer bargaining problem is formulated to find the optimal cut layer between devices and the server. To solve the problem, the Kalai-Smorodinsky bargaining solution (KSBS) is obtained using the bisection method with the feasibility test. Simulation results show that the proposed personalized SL framework with the cut layer from the KSBS can achieve the optimal sum utilities by balancing the energy consumption, training time, and data privacy, and it is also robust to non-iid datasets.

cs.GT↗
arXiv2022

WALLABY Pilot Survey: HI gas kinematics of galaxy pairs in cluster environment

We examine the HI gas kinematics of galaxy pairs in two clusters and a group using Australian Square Kilometre Array Pathfinder (ASKAP) WALLABY pilot survey observations. We compare the HI properties of galaxy pair candidates in the Hydra I and Norma clusters, and the NGC 4636 group, with those of non-paired control galaxies selected in the same fields. We perform HI profile decomposition of the sample galaxies using a tool, {\sc baygaud} which allows us to de-blend a line-of-sight velocity profile with an optimal number of Gaussian components. We construct HI super-profiles of the sample galaxies via stacking of their line profiles after aligning the central velocities. We fit a double Gaussian model to the super-profiles and classify them as kinematically narrow and broad components with respect to their velocity dispersions. Additionally, we investigate the gravitational instability of HI gas disks of the sample galaxies using Toomre Q parameters and HI morphological disturbances. We investigate the effect of the cluster environment on the HI properties of galaxy pairs by dividing the cluster environment into three subcluster regions (i.e., outskirts, infalling and central regions). We find that the denser cluster environment (i.e., infalling and central regions) is likely to impact the HI gas properties of galaxies in a way of decreasing the amplitude of the kinematically narrow HI gas ($M_{\rm{narrow}}^{\rm{HI}}$/$M_{\rm{total}}^{\rm{HI}}$), and increasing the Toomre Q values of the infalling and central galaxies. This tendency is likely to be more enhanced for galaxy pairs in the cluster environment.

astro-ph.GA↗
arXiv2022

Global HI Properties of Galaxies via Super-profile Analysis

We present a new method which constructs an HI super-profile of a galaxy which is based on profile decomposition analysis. The decomposed velocity profiles of an HI data cube with an optimal number of Gaussian components are co-added after being aligned in velocity with respect to their centroid velocities. This is compared to the previous approach where no prior profile decomposition is made for the velocity profiles being stacked. The S/N improved super-profile is useful for deriving the galaxy's global HI properties like velocity dispersion and mass from observations which do not provide sufficient surface brightness sensitivity for the galaxy. As a practical test, we apply our new method to 64 high-resolution HI data cubes of nearby galaxies in the local Universe which are taken from THINGS and LITTLE THINGS. In addition, we also construct two additional HI super-profiles of the sample galaxies using symmetric and all velocity profiles of the cubes whose centroid velocities are determined from Hermite $h_3$ polynomial fitting, respectively. We find that the HI super-profiles constructed using the new method have narrower cores and broader wings in shape than the other two super-profiles. This is mainly due to the effect of either asymmetric velocity profiles' central velocity bias or the removal of asymmetric velocity profiles in the previous methods on the resulting HI super-profiles. We discuss how the shapes ($σ_{\rm{n}}/σ_{\rm{b}}$, $A_{\rm{n}}/A_{\rm{b}}$, and $A_{\rm{n}}/A_{\rm{tot}}$) of the new HI super-profiles which are measured from a double Gaussian fit are correlated with star formation rates of the sample galaxies and are compared with those of the other two super-profiles.

astro-ph.GA↗
arXiv2022

SyncTalkFace: Talking Face Generation with Precise Lip-Syncing via Audio-Lip Memory

The challenge of talking face generation from speech lies in aligning two different modal information, audio and video, such that the mouth region corresponds to input audio. Previous methods either exploit audio-visual representation learning or leverage intermediate structural information such as landmarks and 3D models. However, they struggle to synthesize fine details of the lips varying at the phoneme level as they do not sufficiently provide visual information of the lips at the video synthesis step. To overcome this limitation, our work proposes Audio-Lip Memory that brings in visual information of the mouth region corresponding to input audio and enforces fine-grained audio-visual coherence. It stores lip motion features from sequential ground truth images in the value memory and aligns them with corresponding audio features so that they can be retrieved using audio input at inference time. Therefore, using the retrieved lip motion features as visual hints, it can easily correlate audio with visual dynamics in the synthesis step. By analyzing the memory, we demonstrate that unique lip features are stored in each memory slot at the phoneme level, capturing subtle lip motion based on memory addressing. In addition, we introduce visual-visual synchronization loss which can enhance lip-syncing performance when used along with audio-visual synchronization loss in our model. Extensive experiments are performed to verify that our method generates high-quality video with mouth shapes that best align with the input audio, outperforming previous state-of-the-art methods.

cs.CV↗
arXiv2022

Meta Input: How to Leverage Off-the-Shelf Deep Neural Networks

These days, although deep neural networks (DNNs) have achieved a noticeable progress in a wide range of research area, it lacks the adaptability to be employed in the real-world applications because of the environment discrepancy problem. Such a problem originates from the difference between training and testing environments, and it is widely known that it causes serious performance degradation, when a pretrained DNN model is applied to a new testing environment. Therefore, in this paper, we introduce a novel approach that allows end-users to exploit pretrained DNN models in their own testing environment without modifying the models. To this end, we present a \textit{meta input} which is an additional input transforming the distribution of testing data to be aligned with that of training data. The proposed meta input can be optimized with a small number of testing data only by considering the relation between testing input data and its output prediction. Also, it does not require any knowledge of the network's internal architecture and modification of its weight parameters. Then, the obtained meta input is added to testing data in order to shift the distribution of testing data to that of originally used training data. As a result, end-users can exploit well-trained models in their own testing environment which can differ from the training environment. We validate the effectiveness and versatility of the proposed meta input by showing the robustness against the environment discrepancy through the comprehensive experiments with various tasks.

cs.LG↗
arXiv2022

Transformer Network-based Reinforcement Learning Method for Power Distribution Network (PDN) Optimization of High Bandwidth Memory (HBM)

In this article, for the first time, we propose a transformer network-based reinforcement learning (RL) method for power distribution network (PDN) optimization of high bandwidth memory (HBM). The proposed method can provide an optimal decoupling capacitor (decap) design to maximize the reduction of PDN self- and transfer impedance seen at multiple ports. An attention-based transformer network is implemented to directly parameterize decap optimization policy. The optimality performance is significantly improved since the attention mechanism has powerful expression to explore massive combinatorial space for decap assignments. Moreover, it can capture sequential relationships between the decap assignments. The computing time for optimization is dramatically reduced due to the reusable network on positions of probing ports and decap assignment candidates. This is because the transformer network has a context embedding process to capture meta-features including probing ports positions. In addition, the network is trained with randomly generated data sets. Therefore, without additional training, the trained network can solve new decap optimization problems. The computing time for training and data cost are critically decreased due to the scalability of the network. Thanks to its shared weight property, the network can adapt to a larger scale of problems without additional training. For verification, we compare the results with conventional genetic algorithm (GA), random search (RS), and all the previous RL-based methods. As a result, the proposed method outperforms in all the following aspects: optimality performance, computing time, and data efficiency.

cs.LG↗