skip to main content


Search for: All records

Creators/Authors contains: "Zhang, Wei"

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Word vector embeddings have been shown to contain and amplify biases in the data they are extracted from. Consequently, many techniques have been proposed to identify, mitigate, and attenuate these biases in word representations. In this paper, we utilize interactive visualization to increase the interpretability and accessibility of a collection of state-of-the-art debiasing techniques. To aid this, we present the Visualization of Embedding Representations for deBiasing (“VERB”) system, an open-source web-based visualization tool that helps users gain a technical understanding and visual intuition of the inner workings of debiasing techniques, with a focus on their geometric properties. In particular, VERB offers easy-to-follow examples that explore the effects of these debiasing techniques on the geometry of high-dimensional word vectors. To help understand how various debiasing techniques change the underlying geometry, VERB decomposes each technique into interpretable sequences of primitive transformations and highlights their effect on the word vectors using dimensionality reduction and interactive visual exploration. VERB is designed to target natural language processing (NLP) practitioners who are designing decision-making systems on top of word embeddings, and also researchers working with the fairness and ethics of machine learning systems in NLP. It can also serve as a visual medium for education, which helps an NLP novice understand and mitigate biases in word embeddings. 
    more » « less
    Free, publicly-accessible full text available January 1, 2025
  2. Free, publicly-accessible full text available December 15, 2024
  3. This work presents a hierarchical framework for bipedal locomotion that combines a Reinforcement Learning (RL)-based high-level (HL) planner policy for the online generation of task space commands with a model-based low-level (LL) controller to track the desired task space trajectories. Different from traditional end-to-end learning approaches, our HL policy takes insights from the angular momentum-based linear inverted pendulum (ALIP) to carefully design the observation and action spaces of the Markov Decision Process (MDP). This simple yet effective design creates an insightful mapping between a low-dimensional state that effectively captures the complex dynamics of bipedal locomotion and a set of task space outputs that shape the walking gait of the robot. The HL policy is agnostic to the task space LL controller, which increases the flexibility of the design and generalization of the framework to other bipedal robots. This hierarchical design results in a learning-based framework with improved performance, data efficiency, and robustness compared with the ALIP model-based approach and state-of-the-art learning-based frameworks for bipedal locomotion. The proposed hierarchical controller is tested in three different robots, Rabbit, a five-link underactuated planar biped; Walker2D, a seven-link fully-actuated planar biped; and Digit, a 3D humanoid robot with 20 actuated joints. The trained policy naturally learns human-like locomotion behaviors and is able to effectively track a wide range of walking speeds while preserving the robustness and stability of the walking gait even under adversarial conditions. 
    more » « less
    Free, publicly-accessible full text available October 1, 2024
  4. Free, publicly-accessible full text available December 1, 2024
  5. Flow separation caused by steep topography remains a significant obstacle in accurately predicting turbulent boundary-layer flows over complex terrain, despite the utilization of sophisticated numerical models. The addition of atmospheric thermal stability, in conjunction with steep topography, further complicates the determination of disrupted turbulent wind patterns. The turbulent separated flows over a two-dimensional (2D) steep hill under thermal stratification has not been extensively addressed in previous experimental studies. Such measurements are crucial for enhancing our comprehension of flow physics and validating numerical models. We measured the turbulent wind flows over a 2D steep hill immersed in a stable boundary layer (of the bulk Richardson Number = 0.256) in a thermally-stratified boundary-layer wind tunnel. The flow separation, re-circulation zone and flow reattachment were characterized by the planar particle image velocimetry technique. Vertical profiles of mean air temperature and its fluctuations are also quantified at representative locations above the 2D steep hill and in the near wake region. Results indicate that the separated shear layer, initiated near the crest of the 2D steep hill, dominates the physical process leading to high turbulence levels and the turbulent kinetic energy production in the wake region for both stable and neutral thermal stability. Although the stable boundary layer does not dramatically change the turbulent flow pattern around the hill, the mean separation bubble is elongated by 13%, and its vertical extent is decreased by approximately 20%. Furthermore, the reduced turbulence intensities and turbulent kinetic energy of the near wake flow are attributed to the relatively low turbulence intensity and low momentum of the stable boundary layer due to buoyancy damping, compared to the neutral boundary layer. Additionally, a distinct low-temperature region—a cold pool—is extended beyond the separation bubble, reflecting the significant sheltering effect of the 2D steep hill on the downwind flow and temperature field. 
    more » « less
    Free, publicly-accessible full text available September 1, 2024
  6. Abstract Direct ethanol fuel cells have been widely investigated as nontoxic and low-corrosive energy conversion devices with high energy and power densities. It is still challenging to develop high-activity and durable catalysts for a complete ethanol oxidation reaction on the anode and accelerated oxygen reduction reaction on the cathode. The materials’ physics and chemistry at the catalytic interface play a vital role in determining the overall performance of the catalysts. Herein, we propose a Pd/Co@N-C catalyst that can be used as a model system to study the synergism and engineering at the solid-solid interface. Particularly, the transformation of amorphous carbon to highly graphitic carbon promoted by cobalt nanoparticles helps achieve the spatial confinement effect, which prevents structural degradation of the catalysts. The strong catalyst-support and electronic effects at the interface between palladium and Co@N-C endow the electron-deficient state of palladium, which enhances the electron transfer and improved activity/durability. The Pd/Co@N-C delivers a maximum power density of 438 mW cm −2 in direct ethanol fuel cells and can be operated stably for more than 1000 hours. This work presents a strategy for the ingenious catalyst structural design that will promote the development of fuel cells and other sustainable energy-related technologies. 
    more » « less
    Free, publicly-accessible full text available December 1, 2024
  7. Understanding aqueous outflow resistance at the level of aqueous veins has been a challenge to the management of glaucoma. This study investigated resolving the anatomies of aqueous veins and the textures of surrounding sclera using photoacoustic microscopy (PAM). A dual wavelength PAM system was established and validated using imaging phantoms, porcine and human globes perfused with an optical contrast agentex vivo. The system shows lateral resolution of 8.23 µm and 4.70 µm at 1200 nm and 532 nm, respectively, and an axial resolution of 27.6 µm. The system is able to separately distinguish the aqueous veins and the sclera with high contrast in full circumference of the porcine and human globes.

     
    more » « less
  8. Free, publicly-accessible full text available July 29, 2024
  9. This article studies the class of scenario-based safety testing algorithms in the black-box safety testing configuration. For algorithms sharing the same state–action set coverage with different sampling distributions, it is commonly believed that prioritizing the exploration of high-risk states and actions leads to a better sampling efficiency. Our proposal disputes the above intuition by introducing an impossibility theorem that provably shows that all the safety testing algorithms of the aforementioned difference perform equally well with the same expected sampling efficiency. Moreover, for testing algorithms covering different sets of states and actions, the sampling efficiency criterion is no longer applicable as different algorithms do not necessarily converge to the same termination condition. We then propose a testing aggressiveness definition based on the almost safe set concept along with an unbiased and efficient algorithm that compares the aggressiveness between testing algorithms. Empirical observations from the safety testing of bipedal locomotion controllers and vehicle decision-making modules are also presented to support the proposed theoretical implications and methodologies. 
    more » « less
    Free, publicly-accessible full text available August 1, 2024
  10. Free, publicly-accessible full text available July 29, 2024