Barnabás Póczos
YOU?
Author Swipe
View article: SOFT METROPOLIS-HASTINGS CORRECTION FOR GENERATIVE MODEL SAMPLING
SOFT METROPOLIS-HASTINGS CORRECTION FOR GENERATIVE MODEL SAMPLING Open
Molecular diffusion models suffer from systematic sampling biases that pre- vent optimal structure formation, resulting in chemically suboptimal molecules with metastable conformations trapped in local energy minima. We introduce Metropoli…
View article: Bayesian machine learning for inverse design of ultra-high-performance concrete
Bayesian machine learning for inverse design of ultra-high-performance concrete Open
The diversity of available material feedstocks, coupled with rigorous performance requirements, complicates the design of ultra-high-performance concrete (UHPC). Here, a Bayesian method for inverse design is first demonstrated from publish…
View article: Learning from B Cell Evolution: Adaptive Multi-Expert Diffusion for Antibody Design via Online Optimization
Learning from B Cell Evolution: Adaptive Multi-Expert Diffusion for Antibody Design via Online Optimization Open
Recent advances in diffusion models have shown remarkable potential for antibody design, yet existing approaches apply uniform generation strategies that cannot adapt to each antigen’s unique requirements. Inspired by B cell affinity matur…
View article: Reconstructing Galaxy Cluster Mass Maps using Score-based Generative Modeling
Reconstructing Galaxy Cluster Mass Maps using Score-based Generative Modeling Open
We present a novel approach to reconstruct gas and dark matter projected density maps of galaxy clusters using score-based generative modeling. Our diffusion model takes in mock SZ and X-ray images as conditional inputs, and generates real…
View article: AmpLyze: A Deep Learning Model for Predicting the Hemolytic Concentration
AmpLyze: A Deep Learning Model for Predicting the Hemolytic Concentration Open
Red-blood-cell lysis (HC50) is the principal safety barrier for antimicrobial-peptide (AMP) therapeutics, yet existing models only say "toxic" or "non-toxic." AmpLyze closes this gap by predicting the actual HC50 value from sequence alone …
View article: Recovering time-varying networks from single-cell data
Recovering time-varying networks from single-cell data Open
Motivation Gene regulation is a dynamic process that underlies all aspects of human development, disease response, and other biological processes. The reconstruction of temporal gene regulatory networks has conventionally relied on regress…
View article: Pharmacophore-Conditioned Diffusion Model for Ligand-Based De Novo Drug Design
Pharmacophore-Conditioned Diffusion Model for Ligand-Based De Novo Drug Design Open
Developing bioactive molecules remains a central, time- and cost-heavy challenge in drug discovery, particularly for novel targets lacking structural or functional data. Pharmacophore modeling presents an alternative for capturing the key …
View article: SenSet, a novel human lung senescence cell gene signature, identifies cell-specific senescence mechanisms
SenSet, a novel human lung senescence cell gene signature, identifies cell-specific senescence mechanisms Open
Cellular senescence is a major hallmark of aging. Senescence is defined as an irreversible growth arrest observed when cells are exposed to a variety of stressors including DNA damage, oxidative stress, or nutrient deprivation. While senes…
View article: Chemistry-Inspired Diffusion with Non-Differentiable Guidance
Chemistry-Inspired Diffusion with Non-Differentiable Guidance Open
Recent advances in diffusion models have shown remarkable potential in the conditional generation of novel molecules. These models can be guided in two ways: (i) explicitly, through additional features representing the condition, or (ii) i…
View article: Recovering Time-Varying Networks From Single-Cell Data
Recovering Time-Varying Networks From Single-Cell Data Open
Gene regulation is a dynamic process that underlies all aspects of human development, disease response, and other key biological processes. The reconstruction of temporal gene regulatory networks has conventionally relied on regression ana…
View article: Diffusion Models in De Novo Drug Design
Diffusion Models in De Novo Drug Design Open
Diffusion models have emerged as powerful tools for molecular generation, particularly in the context of 3D molecular structures. Inspired by nonequilibrium statistical physics, these models can generate 3D molecular structures with specif…
View article: GraphBPE: Molecular Graphs Meet Byte-Pair Encoding
GraphBPE: Molecular Graphs Meet Byte-Pair Encoding Open
With the increasing attention to molecular machine learning, various innovations have been made in designing better models or proposing more comprehensive benchmarks. However, less is studied on the data preprocessing schedule for molecula…
View article: Greener GRASS: Enhancing GNNs with Encoding, Rewiring, and Attention
Greener GRASS: Enhancing GNNs with Encoding, Rewiring, and Attention Open
Graph Neural Networks (GNNs) have become important tools for machine learning on graph-structured data. In this paper, we explore the synergistic combination of graph encoding, graph rewiring, and graph attention, by introducing Graph Atte…
View article: Diffusion Models in $\textit{De Novo}$ Drug Design
Diffusion Models in $\textit{De Novo}$ Drug Design Open
Diffusion models have emerged as powerful tools for molecular generation, particularly in the context of 3D molecular structures. Inspired by non-equilibrium statistical physics, these models can generate 3D molecular structures with speci…
View article: Controllable Text Generation in the Instruction-Tuning Era
Controllable Text Generation in the Instruction-Tuning Era Open
While most research on controllable text generation has focused on steering base Language Models, the emerging instruction-tuning and prompting paradigm offers an alternate approach to controllability. We compile and release ConGenBench, a…
View article: Subject2Vec: Generative-Discriminative Approach from a Set of Image Patches to a Vector
Subject2Vec: Generative-Discriminative Approach from a Set of Image Patches to a Vector Open
We propose an attention-based method that aggregates local image features to a subject-level representation for predicting disease severity. In contrast to classical deep learning that requires a fixed dimensional input, our method operate…
View article: Task-Based MoE for Multitask Multilingual Machine Translation
Task-Based MoE for Multitask Multilingual Machine Translation Open
Mixture-of-experts (MoE) architecture has been proven a powerful method for diverse tasks in training deep models in many applications. However, current MoE implementations are task agnostic, treating all tokens from different tasks in the…
View article: Objective-Agnostic Enhancement of Molecule Properties via Multi-Stage VAE
Objective-Agnostic Enhancement of Molecule Properties via Multi-Stage VAE Open
Variational autoencoder (VAE) is a popular method for drug discovery and various architectures and pipelines have been proposed to improve its performance. However, VAE approaches are known to suffer from poor manifold recovery when the da…
View article: SciFix: Outperforming GPT3 on Scientific Factual Error Correction
SciFix: Outperforming GPT3 on Scientific Factual Error Correction Open
Due to the prohibitively high cost of creating error correction datasets, most Factual Claim Correction methods rely on a powerful verification model to guide the correction process. This leads to a significant drop in performance in domai…
View article: Integrated Design of Ultradurable, Low CO<sub>2</sub> Alternative Binder Systems via Machine Learning
Integrated Design of Ultradurable, Low CO<sub>2</sub> Alternative Binder Systems via Machine Learning Open
This ARPA-E project developed a machine learning tool to use in formulation design of cementitious binders for concrete having 50% less embodied CO2 and possessing twice the durability compared to concrete based on ordinary portland cement…
View article: The student becomes the master: Outperforming GPT3 on Scientific Factual Error Correction
The student becomes the master: Outperforming GPT3 on Scientific Factual Error Correction Open
Due to the prohibitively high cost of creating error correction datasets, most Factual Claim Correction methods rely on a powerful verification model to guide the correction process. This leads to a significant drop in performance in domai…
View article: Task-Based MoE for Multitask Multilingual Machine Translation
Task-Based MoE for Multitask Multilingual Machine Translation Open
Mixture-of-experts (MoE) architecture has been proven a powerful method for diverse tasks in training deep models in many applications.However, current MoE implementations are task agnostic, treating all tokens from different tasks in the …
View article: Improving Molecule Properties Through 2-Stage VAE
Improving Molecule Properties Through 2-Stage VAE Open
Variational autoencoder (VAE) is a popular method for drug discovery and there had been a great deal of architectures and pipelines proposed to improve its performance. But the VAE model itself suffers from deficiencies such as poor manifo…
View article: On the Algorithmic Stability and Generalization of Adaptive Optimization Methods
On the Algorithmic Stability and Generalization of Adaptive Optimization Methods Open
Despite their popularity in deep learning and machine learning in general, the theoretical properties of adaptive optimizers such as Adagrad, RMSProp, Adam or AdamW are not yet fully understood. In this paper, we develop a novel framework …
View article: Multiset multicover methods for discriminative marker selection
Multiset multicover methods for discriminative marker selection Open
Markers are increasingly being used for several high-throughput data analysis and experimental design tasks. Examples include the use of markers for assigning cell types in scRNA-seq studies, for deconvolving bulk gene expression data, and…
View article: Coarse-to-Fine Curriculum Learning
Coarse-to-Fine Curriculum Learning Open
When faced with learning challenging new tasks, humans often follow sequences of steps that allow them to incrementally build up the necessary skills for performing these new tasks. However, in machine learning, models are most often train…
View article: Re-TACRED: Addressing Shortcomings of the TACRED Dataset
Re-TACRED: Addressing Shortcomings of the TACRED Dataset Open
TACRED is one of the largest and most widely used sentence-level relation extraction datasets. Proposed models that are evaluated using this dataset consistently set new state-of-the-art performance. However, they still exhibit large error…
View article: Deep generative models for galaxy image simulations
Deep generative models for galaxy image simulations Open
Image simulations are essential tools for preparing and validating the analysis of current and future wide-field optical surveys. However, the galaxy models used as the basis for these simulations are typically limited to simple parametric…
View article: StylePTB: A Compositional Benchmark for Fine-grained Controllable Text\n Style Transfer
StylePTB: A Compositional Benchmark for Fine-grained Controllable Text\n Style Transfer Open
Text style transfer aims to controllably generate text with targeted\nstylistic changes while maintaining core meaning from the source sentence\nconstant. Many of the existing style transfer benchmarks primarily focus on\nindividual high-l…
View article: StylePTB: A Compositional Benchmark for Fine-grained Controllable Text Style Transfer
StylePTB: A Compositional Benchmark for Fine-grained Controllable Text Style Transfer Open
Text style transfer aims to controllably generate text with targeted stylistic changes while maintaining core meaning from the source sentence constant. Many of the existing style transfer benchmarks primarily focus on individual high-leve…