Tony Chiang
YOU?
Author Swipe
View article: Understanding generative AI output with embedding models
Understanding generative AI output with embedding models Open
Constructing high-quality features is critical to any quantitative data analysis. While feature engineering was historically addressed by carefully handcrafting data representations on the basis of domain expertise, deep neural networks (D…
View article: Signatures of quantum spin liquid state and unconventional transport in thin film TbInO3
Signatures of quantum spin liquid state and unconventional transport in thin film TbInO3 Open
Quantum spin liquids, where the frustrated magnetic ground state hosts highly entangled spins resisting long-range order to 0 K, are exotic quantum magnets proximate to unconventional superconductivity and candidate platforms for topologic…
View article: Intertwined polar, chiral, and ferro-rotational orders in a rotation-only insulator
Intertwined polar, chiral, and ferro-rotational orders in a rotation-only insulator Open
Intertwined orders refer to strongly coupled and mutually dependent orders that coexist in correlated electron systems, often underpinning key physical properties of the host materials. Among them, polar, chiral, and ferro-rotational order…
View article: Material-Limited Switching in Nanoscale Ferroelectrics
Material-Limited Switching in Nanoscale Ferroelectrics Open
The ferroelectric switching speed has been experimentally obfuscated by the interaction between the measurement circuit and the ferroelectric switching itself. This has prohibited the observation of real material responses at nanosecond ti…
View article: Assessing Generative Models for Structured Data
Assessing Generative Models for Structured Data Open
Synthetic tabular data generation has emerged as a promising method to address limited data availability and privacy concerns. With the sharp increase in the performance of large language models in recent years, researchers have been inter…
View article: Conductive filament formation in the failure of Hf0.5Zr0.5O2 ferroelectric capacitors
Conductive filament formation in the failure of Hf0.5Zr0.5O2 ferroelectric capacitors Open
Ferroelectric materials provide pathways to higher performance logic and memory technologies, with Hf0.5Zr0.5O2 being the most popular among them. However, critical challenges exist in understanding the material’s failure mechanisms to des…
View article: Thermodynamic origin of nonvolatility in resistive memory
Thermodynamic origin of nonvolatility in resistive memory Open
View article: Understanding Generative AI Content with Embedding Models
Understanding Generative AI Content with Embedding Models Open
Constructing high-quality features is critical to any quantitative data analysis. While feature engineering was historically addressed by carefully hand-crafting data representations based on domain expertise, deep neural networks (DNNs) n…
View article: Measuring training variability from stochastic optimization using robust nonparametric testing
Measuring training variability from stochastic optimization using robust nonparametric testing Open
Deep neural network training often involves stochastic optimization, meaning each run will produce a different model. This implies that hyperparameters of the training process, such as the random seed itself, can potentially have significa…
View article: Nernst coefficient of Pt by non-local electrical measurement
Nernst coefficient of Pt by non-local electrical measurement Open
The Nernst effect describes a linear relationship between orthogonal components of a magnetic field, a temperature gradient, and a resulting transverse electric field. A non-local electrical measurement, where injection and detection are p…
View article: High temperature stability of entropy-stabilized oxide (MgCoNiCuZn)0.2O in air
High temperature stability of entropy-stabilized oxide (MgCoNiCuZn)0.2O in air Open
Entropy-stabilized oxides are single-phase, multicomponent oxides that are stabilized by a large entropy of mixing, ΔS, overcoming a positive enthalpy. Due to the −TΔS term in the Gibbs' free energy, G, it can be hypothesized that entropy-…
View article: Computational and Systems Biology Advances to Enable Bioagent Agnostic Signatures
Computational and Systems Biology Advances to Enable Bioagent Agnostic Signatures Open
Enumerated threat agent lists have long driven biodefense priorities. The global SARS-CoV-2 pandemic demonstrated the limitations of searching for known threat agents as compared to a more agnostic approach. Recent technological advances a…
View article: Endotaxial stabilization of 2D charge density waves with long-range order
Endotaxial stabilization of 2D charge density waves with long-range order Open
Charge density waves are emergent quantum states that spontaneously reduce crystal symmetry, drive metal-insulator transitions, and precede superconductivity. In low-dimensions, distinct quantum states arise, however, thermal fluctuations …
View article: Understanding Generative AI Content with Embedding Models
Understanding Generative AI Content with Embedding Models Open
The construction of high-quality numerical features is critical to any quantitative data analysis. Feature engineering has been historically addressed by carefully hand-crafting data representations based on domain expertise. This work vie…
View article: Report on Pure Mathematics of Machine Learning
Report on Pure Mathematics of Machine Learning Open
We report on the findings conducted by the Pure Mathematics in Machine Learning (PMML) Project under the umbrella of the Mathematics in Artificial Reasoning in Sciences (MARS) Laboratory Initiative at the Pacific Northwest National Laborat…
View article: Efficient kernel surrogates for neural network-based regression
Efficient kernel surrogates for neural network-based regression Open
Despite their immense promise in performing a variety of learning tasks, a theoretical understanding of the limitations of Deep Neural Networks (DNNs) has so far eluded practitioners. This is partly due to the inability to determine the cl…
View article: Computational and Systems Biology Advances to Enable Bioagent-Agnostic Signatures
Computational and Systems Biology Advances to Enable Bioagent-Agnostic Signatures Open
Enumerated threat agent lists have long driven biodefense priorities. The global SARS-CoV-2 pandemic demonstrated the limitations of searching for known threat agents as compared to a more agnostic approach. Recent technological advances a…
View article: Foundation Model's Embedded Representations May Detect Distribution Shift
Foundation Model's Embedded Representations May Detect Distribution Shift Open
Sampling biases can cause distribution shifts between train and test datasets for supervised learning tasks, obscuring our ability to understand the generalization capacity of a model. This is especially important considering the wide adop…
View article: Robust Nonparametric Hypothesis Testing to Understand Variability in Training Neural Networks
Robust Nonparametric Hypothesis Testing to Understand Variability in Training Neural Networks Open
Training a deep neural network (DNN) often involves stochastic optimization, which means each run will produce a different model. Several works suggest this variability is negligible when models have the same performance, which in the case…
View article: Exploring Learned Representations of Neural Networks with Principal Component Analysis
Exploring Learned Representations of Neural Networks with Principal Component Analysis Open
Understanding feature representation for deep neural networks (DNNs) remains an open question within the general field of explainable AI. We use principal component analysis (PCA) to study the performance of a k-nearest neighbors classifie…
View article: Minibatching Offers Improved Generalization Performance for Second Order Optimizers
Minibatching Offers Improved Generalization Performance for Second Order Optimizers Open
Training deep neural networks (DNNs) used in modern machine learning is computationally expensive. Machine learning scientists, therefore, rely on stochastic first-order methods for training, coupled with significant hand-tuning, to obtain…
View article: Faithful and Efficient Explanations for Neural Networks via Neural Tangent Kernel Surrogate Models
Faithful and Efficient Explanations for Neural Networks via Neural Tangent Kernel Surrogate Models Open
A recent trend in explainable AI research has focused on surrogate modeling, where neural networks are approximated as simpler ML algorithms such as kernel machines. A second trend has been to utilize kernel functions in various explain-by…
View article: Data for EMSL Project 51834 from January 2023
Data for EMSL Project 51834 from January 2023 Open
View article: Data for EMSL Project 51834 from January 2023
Data for EMSL Project 51834 from January 2023 Open
View article: pnnl/projection_ntk
pnnl/projection_ntk Open
This is a repository for work conducted in our research paper we would like to make open source. The main features are that we forked another open source library and added functionality that enabled us to compute neural-kernel objects usin…
View article: Data for EMSL Project 51834 from January 2023
Data for EMSL Project 51834 from January 2023 Open
View article: Spectral Evolution and Invariance in Linear-width Neural Networks
Spectral Evolution and Invariance in Linear-width Neural Networks Open
We investigate the spectral properties of linear-width feed-forward neural networks, where the sample size is asymptotically proportional to network width. Empirically, we show that the spectra of weight in this high dimensional regime are…
View article: TorchNTK: A Library for Calculation of Neural Tangent Kernels of PyTorch Models
TorchNTK: A Library for Calculation of Neural Tangent Kernels of PyTorch Models Open
We introduce torchNTK, a python library to calculate the empirical neural tangent kernel (NTK) of neural network models in the PyTorch framework. We provide an efficient method to calculate the NTK of multilayer perceptrons. We compare the…
View article: Data for EMSL Project 51834 from October 2022
Data for EMSL Project 51834 from October 2022 Open
View article: Data for EMSL Project 51834 from September 2022
Data for EMSL Project 51834 from September 2022 Open