Esmeralda Whitammer
PulseAugur coverage of Esmeralda Whitammer — every cluster mentioning Esmeralda Whitammer across labs, papers, and developer communities, ranked by signal.
-
PhyloGFN uses generative flow networks for evolutionary relationship inference
Researchers have developed PhyloGFN, a novel method for phylogenetic inference that utilizes generative flow networks (GFlowNets). This approach is designed to address the computational challenges in reconstructing evol…
-
New method models stochastic dynamics without data samples
Researchers have developed a novel method for modeling Schrödinger bridges, which are stochastic dynamical systems that connect two probability distributions. This new approach, termed data-to-energy IPF, allows for the…
-
New RL algorithm TGM enhances candidate filtering for scientific discovery
Researchers have developed a new method for scientific discovery that aims to improve the filtering of potential candidates like proteins or molecules. This approach uses a unified operator that combines several regular…
-
New adaptive teacher method improves amortized inference for complex distributions
Researchers have developed a novel approach to amortized inference, a process used to train parametric models like neural networks for approximating complex distributions. This new method utilizes an adaptive training d…
-
New flow networks tackle stochastic and adversarial AI challenges
Researchers have introduced Expected Flow Networks (EFlowNets), an advancement upon Generative Flow Networks (GFlowNets), designed to operate effectively in stochastic environments. These EFlowNets have demonstrated sup…
-
New Delta-AI algorithm speeds up inference in sparse graphical models
Researchers have introduced Delta-AI, a novel algorithm designed for amortized inference in sparse probabilistic graphical models (PGMs). This method leverages the sparsity of PGMs to enable local credit assignment with…
-
GFlowNets and variational inference algorithms shown to be equivalent in certain cases
This paper explores the connection between Generative Flow Networks (GFlowNets) and variational inference (VI), two families of probabilistic algorithms. The authors demonstrate that VI algorithms can be seen as special…
-
GFlowNets enhance LLM red-teaming for improved AI safety
Researchers have developed a new method using GFlowNets to improve the diversity and effectiveness of automated red-teaming for large language models (LLMs). This approach aims to discover a wider range of harmful promp…
-
Two arXiv papers review probabilistic generative models and introduce new quality assessment methods
Two new arXiv papers explore advancements in probabilistic generative models. The first, a review, focuses on Probabilistic Circuits (PCs) and their trade-offs between expressivity and tractability, discussing hybrid de…
-
New flow matching methods enhance generative models for design and imaging · 6 sources tracked
Researchers are exploring advanced flow matching techniques to enhance generative models for inverse design problems and image generation. Conditional Flow Matching (CFM) shows promise in engineering inverse design, out…
-
Diffusion models show inherent attention mechanisms and improved sampling techniques · 7 sources tracked
Recent research explores advancements in diffusion models, a dominant architecture for image generation. One paper reveals that these models inherently utilize an attention mechanism similar to transformers, suggesting …
-
New research improves diffusion model training with novel sampling strategy
A new research paper explores methods for training diffusion models to sample from distributions defined by unnormalized densities or energy functions. The study benchmarks existing simulation-based variational and off-…
-
New GFlowNet method jointly infers Bayesian Network structure and parameters
Researchers have developed JSP-GFN, a novel method utilizing Generative Flow Networks (GFlowNets) to jointly infer both the structure and parameters of Bayesian Networks. This approach extends existing GFlowNet applicat…
-
GFlowNets applied to solve complex combinatorial optimization problems
Researchers have developed a novel approach using GFlowNets to tackle complex combinatorial optimization problems, which are often too difficult for traditional algorithms. The method involves designing specific Markov …
-
New 'trajectory balance' objective improves GFlowNet learning
Researchers have introduced "trajectory balance," a novel learning objective for Generative Flow Networks (GFlowNets). This new objective aims to improve credit assignment in GFlowNets, which are used for generating com…