arXiv subjects

Tianyu Wang

Publications and source records attributed to Tianyu Wang.

At least 163 records · Page 9Linked to original sources

arXiv2024

Improving GPU Multi-Tenancy Through Dynamic Multi-Instance GPU Reconfiguration

Continuous learning (CL) has emerged as one of the most popular deep learning paradigms deployed in modern cloud GPUs. Specifically, CL has the capability to continuously update the model parameters (through model retraining) and use the updated model (if available) to serve overtime arriving inference requests. It is generally beneficial to co-locate the retraining and inference together to enable timely model updates and avoid model transfer overheads. This brings the need for GPU sharing among retraining and inferences. Meanwhile, multiple CL workloads can share the modern GPUs in the cloud, leading to multi-tenancy execution. In this paper, we observe that prior GPU-sharing techniques are not optimized for multi-tenancy CL workloads. Specifically, they do not coherently consider the accuracy of the retraining model and the inference service level objective (SLO) attainment. Moreover, they cannot accommodate the overtime dynamics (e.g., inference arrival intensity) in CL execution. In this paper, we propose MIGRator, a novel GPU reconfiguration runtime that dynamically performs GPU reconfiguration for multi-tenancy CL workloads. MIGRator is based on the recent NVIDIA multi-instance GPU (MIG) to mitigate resource contention and formulates the reconfiguration optimization into Integer Linear Programming (ILP) to dynamically identify, reconfigure, and allocate the GPU instances. MIGRator leverages the "Goodput" metric in the ILP objective function to consider both inference SLO attainment and model accuracy in the reconfiguration exploration. We evaluate MIGRator using representative multi-tenancy CL workloads. The results show our approach outperforms the state-of-the-art GPU sharing techniques (i.e., Ekya, Astraea, and PARIS) by 17\%, 21\%, and 20\%, respectively.

cs.DC↗

arXiv2024

Enhancing Computer Programming Education with LLMs: A Study on Effective Prompt Engineering for Python Code Generation

Large language models (LLMs) and prompt engineering hold significant potential for advancing computer programming education through personalized instruction. This paper explores this potential by investigating three critical research questions: the systematic categorization of prompt engineering strategies tailored to diverse educational needs, the empowerment of LLMs to solve complex problems beyond their inherent capabilities, and the establishment of a robust framework for evaluating and implementing these strategies. Our methodology involves categorizing programming questions based on educational requirements, applying various prompt engineering strategies, and assessing the effectiveness of LLM-generated responses. Experiments with GPT-4, GPT-4o, Llama3-8b, and Mixtral-8x7b models on datasets such as LeetCode and USACO reveal that GPT-4o consistently outperforms others, particularly with the "multi-step" prompt strategy. The results show that tailored prompt strategies significantly enhance LLM performance, with specific strategies recommended for foundational learning, competition preparation, and advanced problem-solving. This study underscores the crucial role of prompt engineering in maximizing the educational benefits of LLMs. By systematically categorizing and testing these strategies, we provide a comprehensive framework for both educators and students to optimize LLM-based learning experiences. Future research should focus on refining these strategies and addressing current LLM limitations to further enhance educational outcomes in computer programming instruction.

cs.AI↗

arXiv2024

Evaluation of Text-to-Video Generation Models: A Dynamics Perspective

Comprehensive and constructive evaluation protocols play an important role in the development of sophisticated text-to-video (T2V) generation models. Existing evaluation protocols primarily focus on temporal consistency and content continuity, yet largely ignore the dynamics of video content. Dynamics are an essential dimension for measuring the visual vividness and the honesty of video content to text prompts. In this study, we propose an effective evaluation protocol, termed DEVIL, which centers on the dynamics dimension to evaluate T2V models. For this purpose, we establish a new benchmark comprising text prompts that fully reflect multiple dynamics grades, and define a set of dynamics scores corresponding to various temporal granularities to comprehensively evaluate the dynamics of each generated video. Based on the new benchmark and the dynamics scores, we assess T2V models with the design of three metrics: dynamics range, dynamics controllability, and dynamics-based quality. Experiments show that DEVIL achieves a Pearson correlation exceeding 90% with human ratings, demonstrating its potential to advance T2V generation models. Code is available at https://github.com/MingXiangL/DEVIL.

cs.CV↗

arXiv2024

Nuclear mass table in deformed relativistic Hartree-Bogoliubov theory in continuum, II: Even-$Z$ nuclei

The mass table in the deformed relativistic Hartree-Bogoliubov theory in continuum (DRHBc) with the PC-PK1 density functional has been established for even-$Z$ nuclei with $8\le Z\le120$, extended from the previous work for even-even nuclei [Zhang $\it{et.~al.}$ (DRHBc Mass Table Collaboration), At. Data Nucl. Data Tables 144, 101488 (2022)]. The calculated binding energies, two-nucleon and one-neutron separation energies, root-mean-square (rms) radii of neutron, proton, matter, and charge distributions, quadrupole deformations, and neutron and proton Fermi surfaces are tabulated and compared with available experimental data. A total of 4829 even-$Z$ nuclei are predicted to be bound, with an rms deviation of 1.477 MeV from the 1244 mass data. Good agreement with the available experimental odd-even mass differences, $α$ decay energies, and charge radii is also achieved. The description accuracy for nuclear masses and nucleon separation energies as well as the prediction for drip lines is compared with the results obtained from other relativistic and nonrelativistic density functional. The comparison shows that the DRHBc theory with PC-PK1 provides an excellent microscopic description for the masses of even-$Z$ nuclei. The systematics of the nucleon separation energies, odd-even mass differences, pairing energies, two-nucleon gaps, $α$ decay energies, rms radii, quadrupole deformations, potential energy curves, neutron density distributions, and neutron mean-field potentials are discussed.

nucl-th↗

arXiv2024

Cross-Embodiment Robot Manipulation Skill Transfer using Latent Space Alignment

This paper focuses on transferring control policies between robot manipulators with different morphology. While reinforcement learning (RL) methods have shown successful results in robot manipulation tasks, transferring a trained policy from simulation to a real robot or deploying it on a robot with different states, actions, or kinematics is challenging. To achieve cross-embodiment policy transfer, our key insight is to project the state and action spaces of the source and target robots to a common latent space representation. We first introduce encoders and decoders to associate the states and actions of the source robot with a latent space. The encoders, decoders, and a latent space control policy are trained simultaneously using loss functions measuring task performance, latent dynamics consistency, and encoder-decoder ability to reconstruct the original states and actions. To transfer the learned control policy, we only need to train target encoders and decoders that align a new target domain to the latent space. We use generative adversarial training with cycle consistency and latent dynamics losses without access to the task reward or reward tuning in the target domain. We demonstrate sim-to-sim and sim-to-real manipulation policy transfer with source and target robots of different states, actions, and embodiments. The source code is available at \url{https://github.com/ExistentialRobotics/cross_embodiment_transfer}.

cs.RO↗

arXiv2024

Personalized Interiors at Scale: Leveraging AI for Efficient and Customizable Design Solutions

In this paper, we introduce an innovative application of artificial intelligence in the realm of interior design through the integration of Stable Diffusion and Dreambooth models. This paper explores the potential of these advanced generative models to streamline and democratize the process of room interior generation, offering a significant departure from conventional, labor-intensive techniques. Our approach leverages the capabilities of Stable Diffusion for generating high-quality images and Dreambooth for rapid customization with minimal training data, addressing the need for efficiency and personalization in the design industry. We detail a comprehensive methodology that combines these models, providing a robust framework for the creation of tailored room interiors that reflect individual tastes and functional requirements. We presents an extensive evaluation of our method, supported by experimental results that demonstrate its effectiveness and a series of case studies that illustrate its practical application in interior design projects. Our study contributes to the ongoing discourse on the role of AI in creative fields, highlighting the benefits of leveraging generative models to enhance creativity and reshape the future of interior design.

cs.HC↗

arXiv2024

Jet mixing enhancement with Bayesian optimization, deep learning, and persistent data topology

We optimize the jet mixing using large eddy simulations (LES) at a Reynolds number of $3000$. Key methodological enablers consist of Bayesian optimization, a surrogate model enhanced by deep learning, and persistent data topology for physical interpretation. The mixing performance is characterized by an equivalent jet radius ($R_{\rm eq}$) derived from the streamwise velocity in a plane located $8$ diameters downstream. The optimization is performed in a 22-dimensional actuation space that comprises most known excitations. The plant benefits from a 22-dimensional actuation space that comprises most known excitations. This search space parameterizes distributed actuation imposed on the bulk flow and at the periphery of the nozzle in the streamwise and radial directions. The momentum flux measures the energy input of the actuation. The optimization quadruples the jet radius $R_{\rm eq}$ with a $7$-armed blooming jet after around $570$ evaluations. The control input requires $2\%$ momentum flux of the main flow, which is one order of magnitude lower than an ad hoc dual-mode excitation. Intriguingly, a pronounced suboptimum in the search space is associated with a double-helix jet, a new flow pattern. This jet pattern results in a mixing improvement comparable to the blooming jet. A state-of-the-art Bayesian optimization converges towards this double helix solution. The learning is accelerated and converges to another better optimum by including surrogate model trained along the optimization. Persistent data topology extracts the global and many local minima in the actuation space. These minima can be identified with flow patterns beneficial to the mixing.

physics.flu-dyn↗

arXiv2024

Occam Factor for Random Graphs: Erdös-Rényi, Independent Edge, and Rank-1 Stochastic Blockmodel

We investigate the evidence/flexibility (i.e., "Occam") paradigm and demonstrate the theoretical and empirical consistency of Bayesian evidence for the task of determining an appropriate generative model for network data. This model selection framework involves determining a collection of candidate models, equipping each of these models' parameters with prior distributions derived via the encompassing priors method, and computing or approximating each models' evidence. We demonstrate how such a criterion may be used to select the most suitable model among the Erdös-Rényi (ER) model, independent edge (IE) model, and rank-1 stochastic blockmodel (SBM). The Erdös-Rényi may be considered as being linearly nested within IE, a fact which permits exponential family results. The rank-1 SBM is not so ideal, so we propose a numerical method to approximate its evidence. We apply this paradigm to brain connectome data. Future work necessitates deriving and equipping additional candidate random graph models with appropriate priors so they may be included in the paradigm.

stat.ME↗

arXiv2024

Experimental jet control with Bayesian optimization and persistent data topology

This study experimentally optimizes the mixing of a turbulent jet at $Re=10000$ with the surrounding air by targeted shear layer actuation. The forcing is composed of superposed harmonic signals of different azimuthal wavenumber $m$ generated by eight loudspeakers circumferentially distributed around the nozzle lip. Amplitudes and frequencies of the individual harmonic contributions serve as optimization parameters and the time-averaged centerline velocity downstream of the potential core is used as a metric for mixing optimization. The actuation is optimized through Bayesian optimization. Three search spaces are explored - axisymmetric forcing, $m=0$, superposed axisymmetric and helical forcing, $m \in \{0,1\}$, and axisymmetric actuation combined with two counter-rotating helical modes, $m \in \{-1,0,1\}$. High-speed PIV is employed to analyze the jet response to the optimized forcing. The optimization processes are analyzed by persistent data topology. In the search space of axisymmetric excitation, the routine identifies an actuation at the natural frequency of the flow to be most efficient, with the centerline velocity being decreased by $15\%$. The optimal solutions in both the two-mode and three-mode search space converge to a similar forcing with one axial and one helical mode combined at a frequency ratio of around $2.3$. Spectral analysis of the PIV images reveals that for the identified optimal forcing frequencies, a non-linear interaction between forced and natural structures in the jet flow is triggered, leading to a reduction in centerline velocity of around $35\%$. The topology of the most complex search space from the discrete data reveals four basins of attractions, classified into three forcing patterns including axisymmetric, axisym.-helical, and axisym.-flapping. Two deep basins are related to the optimal axisym.-helical pattern, and the others are shallower.

physics.flu-dyn↗

arXiv2024

Improving Multi-Instance GPU Efficiency via Sub-Entry Sharing TLB Design

NVIDIA's Multi-Instance GPU (MIG) technology enables partitioning GPU computing power and memory into separate hardware instances, providing complete isolation including compute resources, caches, and memory. However, prior work identifies that MIG does not extend to partitioning the last-level TLB (i.e., L3 TLB), which remains shared among all instances. To enhance TLB reach, NVIDIA GPUs reorganized the TLB structure with 16 sub-entries in each L3 TLB entry that have a one-to-one mapping to the address translations for 16 pages of size 64KB located within the same 1MB aligned range. Our comprehensive investigation of address translation efficiency in MIG identifies two main issues caused by L3 TLB sharing interference: (i) it results in performance degradation for co-running applications, and (ii) TLB sub-entries are not fully utilized before eviction. Based on this observation, we propose STAR to improve the utilization of TLB sub-entries through dynamic sharing of TLB entries across multiple base addresses. STAR evaluates TLB entries based on their sub-entry utilization to optimize address translation storage, dynamically adjusting between a shared and non-shared status to cater to current demand. We show that STAR improves overall performance by an average of 30.2% across various multi-tenant workloads.

cs.DC↗

arXiv2024

Mining Supervision for Dynamic Regions in Self-Supervised Monocular Depth Estimation

This paper focuses on self-supervised monocular depth estimation in dynamic scenes trained on monocular videos. Existing methods jointly estimate pixel-wise depth and motion, relying mainly on an image reconstruction loss. Dynamic regions1 remain a critical challenge for these methods due to the inherent ambiguity in depth and motion estimation, resulting in inaccurate depth estimation. This paper proposes a self-supervised training framework exploiting pseudo depth labels for dynamic regions from training data. The key contribution of our framework is to decouple depth estimation for static and dynamic regions of images in the training data. We start with an unsupervised depth estimation approach, which provides reliable depth estimates for static regions and motion cues for dynamic regions and allows us to extract moving object information at the instance level. In the next stage, we use an object network to estimate the depth of those moving objects assuming rigid motions. Then, we propose a new scale alignment module to address the scale ambiguity between estimated depths for static and dynamic regions. We can then use the depth labels generated to train an end-to-end depth estimation network and improve its performance. Extensive experiments on the Cityscapes and KITTI datasets show that our self-training strategy consistently outperforms existing self/unsupervised depth estimation methods.

cs.CV↗

arXiv2024

WM-MoE: Weather-aware Multi-scale Mixture-of-Experts for Blind Adverse Weather Removal

Adverse weather removal tasks like deraining, desnowing, and dehazing are usually treated as separate tasks. However, in practical autonomous driving scenarios, the type, intensity,and mixing degree of weather are unknown, so handling each task separately cannot deal with the complex practical scenarios. In this paper, we study the blind adverse weather removal problem. Mixture-of-Experts (MoE) is a popular model that adopts a learnable gate to route the input to different expert networks. The principle of MoE involves using adaptive networks to process different types of unknown inputs. Therefore, MoE has great potential for blind adverse weather removal. However, the original MoE module is inadequate for coupled multiple weather types and fails to utilize multi-scale features for better performance. To this end, we propose a method called Weather-aware Multi-scale MoE (WM-MoE) based on Transformer for blind weather removal. WM-MoE includes two key designs: WEather-Aware Router (WEAR) and Multi-Scale Experts (MSE). WEAR assigns experts for each image token based on decoupled content and weather features, which enhances the model's capability to process multiple adverse weathers. To obtain discriminative weather features from images, we propose Weather Guidance Fine-grained Contrastive Learning (WGF-CL), which utilizes weather cluster information to guide the assignment of positive and negative samples for each image token. Since processing different weather types requires different receptive fields, MSE leverages multi-scale features to enhance the spatial relationship modeling capability, facilitating the high-quality restoration of diverse weather types and intensities. Our method achieves state-of-the-art performance in blind adverse weather removal on two public datasets and our dataset. We also demonstrate the advantage of our method on downstream segmentation tasks.

cs.CV↗

arXiv2024

From Random Search to Bandit Learning in Metric Measure Spaces

Random Search is one of the most widely-used method for Hyperparameter Optimization, and is critical to the success of deep learning models. Despite its astonishing performance, little non-heuristic theory has been developed to describe the underlying working mechanism. This paper gives a theoretical accounting of Random Search. We introduce the concept of \emph{scattering dimension} that describes the landscape of the underlying function, and quantifies the performance of random search. We show that, when the environment is noise-free, the output of random search converges to the optimal value in probability at rate $ \widetilde{\mathcal{O}} \left( \left( \frac{1}{T} \right)^{ \frac{1}{d_s} } \right) $, where $ d_s \ge 0 $ is the scattering dimension of the underlying function. When the observed function values are corrupted by bounded $iid$ noise, the output of random search converges to the optimal value in probability at rate $ \widetilde{\mathcal{O}} \left( \left( \frac{1}{T} \right)^{ \frac{1}{d_s + 1} } \right) $. In addition, based on the principles of random search, we introduce an algorithm, called BLiN-MOS, for Lipschitz bandits in doubling metric spaces that are also endowed with a probability measure, and show that under mild conditions, BLiN-MOS achieves a regret rate of order $ \widetilde{\mathcal{O}} \left( T^{ \frac{d_z}{d_z + 1} } \right) $, where $d_z$ is the zooming dimension of the problem instance.

cs.LG↗

arXiv2024

Zeroth-order Low-rank Hessian Estimation via Matrix Recovery

A zeroth-order Hessian estimator aims to recover the Hessian matrix of an objective function at any given point, using minimal finite-difference computations. This paper studies zeroth-order Hessian estimation for low-rank Hessians, from a matrix recovery perspective. Our challenge lies in the fact that traditional matrix recovery techniques are not directly suitable for our scenario. They either demand incoherence assumptions (or its variants), or require an impractical number of finite-difference computations in our setting. To overcome these hurdles, we employ zeroth-order Hessian estimations aligned with proper matrix measurements, and prove new recovery guarantees for these estimators. More specifically, we prove that for a Hessian matrix $H \in \mathbb{R}^{n \times n}$ of rank $r$, $ \mathcal{O}(nr^2 \log^2 n ) $ proper zeroth-order finite-difference computations ensures a highly probable exact recovery of $H$. Compared to existing methods, our method can greatly reduce the number of finite-difference computations, and does not require any incoherence assumptions.

math.OC↗

arXiv2024

Mechanical Intelligence Simplifies Control in Terrestrial Limbless Locomotion

Limbless locomotors, from microscopic worms to macroscopic snakes, traverse complex, heterogeneous natural environments typically using undulatory body wave propagation. Theoretical and robophysical models typically emphasize body kinematics and active neural/electronic control. However, we contend that because such approaches often neglect the role of passive, mechanically controlled processes (those involving "mechanical intelligence"), they fail to reproduce the performance of even the simplest organisms. To uncover principles of how mechanical intelligence aids limbless locomotion in heterogeneous terradynamic regimes, here we conduct a comparative study of locomotion in a model of heterogeneous terrain (lattices of rigid posts). We used a model biological system, the highly studied nematode worm Caenorhabditis elegans, and a robophysical device whose bilateral actuator morphology models that of limbless organisms across scales. The robot's kinematics quantitatively reproduced the performance of the nematodes with purely open-loop control; mechanical intelligence simplified control of obstacle navigation and exploitation by reducing the need for active sensing and feedback. An active behavior observed in C. elegans, undulatory wave reversal upon head collisions, robustified locomotion via exploitation of the systems' mechanical intelligence. Our study provides insights into how neurally simple limbless organisms like nematodes can leverage mechanical intelligence via appropriately tuned bilateral actuation to locomote in complex environments. These principles likely apply to neurally more sophisticated organisms and also provide a design and control paradigm for limbless robots for applications like search and rescue and planetary exploration.

cs.RO↗

arXiv2023

Probabilistic and Semantic Descriptions of Image Manifolds and Their Applications

This paper begins with a description of methods for estimating image probability density functions that reflects the observation that such data is usually constrained to lie in restricted regions of the high-dimensional image space-not every pattern of pixels is an image. It is common to say that images lie on a lower-dimensional manifold in the high-dimensional space. However, it is not the case that all points on the manifold have an equal probability of being images. Images are unevenly distributed on the manifold, and our task is to devise ways to model this distribution as a probability distribution. We therefore consider popular generative models. For our purposes, generative/probabilistic models should have the properties of 1) sample generation: the possibility to sample from this distribution with the modelled density function, and 2) probability computation: given a previously unseen sample from the dataset of interest, one should be able to compute its probability, at least up to a normalising constant. To this end, we investigate the use of methods such as normalising flow and diffusion models. We then show how semantic interpretations are used to describe points on the manifold. To achieve this, we consider an emergent language framework that uses variational encoders for a disentangled representation of points that reside on a given manifold. Trajectories between points on a manifold can then be described as evolving semantic descriptions. We also show that such probabilistic descriptions (bounded) can be used to improve semantic consistency by constructing defences against adversarial attacks. We evaluate our methods with improved semantic robustness and OoD detection capability, explainable and editable semantic interpolation, and improved classification accuracy under patch attacks. We also discuss the limitation in diffusion models.

cs.CV↗

arXiv2023

Geometry-Calibrated DRO: Combating Over-Pessimism with Free Energy Implications

Machine learning algorithms minimizing average risk are susceptible to distributional shifts. Distributionally Robust Optimization (DRO) addresses this issue by optimizing the worst-case risk within an uncertainty set. However, DRO suffers from over-pessimism, leading to low-confidence predictions, poor parameter estimations as well as poor generalization. In this work, we conduct a theoretical analysis of a probable root cause of over-pessimism: excessive focus on noisy samples. To alleviate the impact of noise, we incorporate data geometry into calibration terms in DRO, resulting in our novel Geometry-Calibrated DRO (GCDRO) for regression. We establish the connection between our risk objective and the Helmholtz free energy in statistical physics, and this free-energy-based risk can extend to standard DRO methods. Leveraging gradient flow in Wasserstein space, we develop an approximate minimax optimization algorithm with a bounded error ratio and elucidate how our approach mitigates noisy sample effects. Comprehensive experiments confirm GCDRO's superiority over conventional DRO methods.

cs.LG↗

arXiv2023

Lipschitz Bandits with Batched Feedback

In this paper, we study Lipschitz bandit problems with batched feedback, where the expected reward is Lipschitz and the reward observations are communicated to the player in batches. We introduce a novel landscape-aware algorithm, called Batched Lipschitz Narrowing (BLiN), that optimally solves this problem. Specifically, we show that for a $T$-step problem with Lipschitz reward of zooming dimension $d_z$, our algorithm achieves theoretically optimal (up to logarithmic factors) regret rate $\widetilde{\mathcal{O}}\left(T^{\frac{d_z+1}{d_z+2}}\right)$ using only $ \mathcal{O} \left( \log\log T\right) $ batches. We also provide complexity analysis for this problem. Our theoretical lower bound implies that $Ω(\log\log T)$ batches are necessary for any algorithm to achieve the optimal regret. Thus, BLiN achieves optimal regret rate (up to logarithmic factors) using minimal communication.

cs.LG↗