Anh Bui

• University of Social Sciences and Humanities

Concept erasure has emerged as a promising technique for mitigating the risk of harmful content generation in diffusion models by selectively unlearning undesirable concepts. The common principle of previous works to remove a specific concept is to map it to a fixed generic concept, such as a neutral concept or just an empty text prompt. In this pa...

Diffusion models excel at generating visually striking content from text but can inadvertently produce undesirable or harmful content when trained on unfiltered internet data. A practical solution is to selectively removing target concepts from the model, but this may impact the remaining concepts. Prior approaches have tried to balance this by int...

Deep learning models, even the-state-of-the-art ones, are highly vulnerable to adversarial examples. Adversarial training is one of the most efficient methods to improve the model's robustness. The key factor for the success of adversarial training is the capability to generate qualified and divergent adversarial examples which satisfy some objecti...

The United States is the third largest electric vehicle manufacturer, behind China and Europe. From 2017 to 2020, the U.S. share of cumulative global electric vehicle production since 2010 decreased from 20% to 18%, while manufacturing increased in the other two regions. The electric vehicle production dynamics reflect where there has been the mos...

This report quantifies Los Angeles electric vehicle charging infrastructure needs and associated energy demand citywide in 2030 to align with the city’s goal of 25% zero-emission vehicle stock by 2025 and 80% by 2035. It also quantifies needs in specific areas that are poised to become zero-emission areas or fully fossil fuel free by 2030. Los Ang...

Contrastive learning (CL) has recently emerged as an effective approach to learning representation in a range of downstream tasks. Central to this approach is the selection of positive (similar) and negative (dissimilar) sets to provide the model the opportunity to `contrast' between data and class representation in the latent space. In this paper,...

The fact that deep neural networks are susceptible to crafted perturbations severely impacts the use of deep learning in certain domains of application. Among many developed defense models against such attacks, adversarial training emerges as the most successful method that consistently resists a wide range of attacks. In this work, based on an obs...

Ensemble-based adversarial training is a principled approach to achieve robustness against adversarial attacks. An important technique of this approach is to control the transferability of adversarial examples among ensemble members. We propose in this work a simple yet effective strategy to collaborate among committee models of an ensemble model....

This briefing builds upon our annual U.S. electric vehicle market analysis of state, local, and utility actions to promote electric vehicles. There are major uncertainties about how the industry and the market will respond to the 2020 pandemic and economic recovery, but this briefing offers a status update on many actions that remain critical to su...

The fact that deep neural networks are susceptible to crafted perturbations severely impacts the use of deep learning in certain domains of application. Among many developed defense models against such attacks, adversarial training emerges as the most successful method that consistently resists a wide range of attacks. In this work, based on an obs...

Purpose: To assess opioid prescribing patterns among American Society of Ophthalmic Plastic and Reconstructive Surgery (ASOPRS) members. Methods: An observational, retrospective, cohort study of ASOPRS members' prescribing patterns in the 2013-2016 Medicare Part D Prescriber database. These prescribers were stratified by years in practice, sex,...

Epilepsy is a complex disorder involving neurological alterations that lead to the pathological development of spontaneous, recurrent seizures. For decades, seizures were thought to be largely repetitive, and had been examined at the macrocircuit level using electrophysiological recordings. However, research mapping the dynamics of large neuronal p...

The boundary value problem (BVP) for the Ablowitz-Ladik (AL) system on the natural numbers with linearizable boundary conditions is studied. In particular: (i) a discrete analogue is derived of the Bcklund transformation that was used to solved similar BVPs for the nonlinear Schrödinger equation; (ii) an explicit proof is given that the Bcklund-tra...

Genetic defects in the UBE3A gene, which encodes for the imprinted E6-AP ubiquitin E3 ligase (UBE3A), is responsible for the occurrence of Angelman syndrome (AS), a neurodegenerative disorder which arises in 1 out of every 12,000–20,000 births. Classical symptoms of AS include delayed development, impaired speech, and epileptic seizures with charac...

The dentate gyrus is a region subject to intense study in epilepsy due to its posited role as a 'gate' - acting to inhibit overexcitation in the hippocampal circuitry through its unique synaptic, cellular, and network properties that result in relatively low excitability. Numerous changes predicted to produce dentate hyperexcitability are seen in e...

Boundary value problems for the nonlinear Schrödinger equations on the half line with homogeneous Robin boundary conditions are revisited using Bäcklund transformations. In particular: relations are obtained among the norming constants associated with symmetric eigenvalues; a linearizing transformation is derived for the Bäcklund transformation; th...