English
Related papers

Related papers: Maximum Entropy Modeling Toolkit

200 papers

In streaming settings, speech recognition models have to map sub-sequences of speech to text before the full audio stream becomes available. However, since alignment information between speech and text is rarely available during training,…

Audio and Speech Processing · Electrical Eng. & Systems 2023-12-20 Oscar Chang , Dongseong Hwang , Olivier Siohan

As Large Language Models (LLMs) are deployed more widely, customization with respect to vocabulary, style, and character becomes more important. In this work, we introduce model arithmetic, a novel inference framework for composing and…

Computation and Language · Computer Science 2024-03-07 Jasper Dekoninck , Marc Fischer , Luca Beurer-Kellner , Martin Vechev

Long-term training of large language models (LLMs) requires maintaining stable exploration to prevent the model from collapsing into sub-optimal behaviors. Entropy is crucial in this context, as it controls exploration and helps avoid…

Machine Learning · Computer Science 2026-02-03 Kai Yang , Xin Xu , Yangkun Chen , Weijie Liu , Jiafei Lyu , Zichuan Lin , Deheng Ye , Saiyong Yang

The paper proposes a new message passing algorithm for cycle-free factor graphs. The proposed "entropy message passing" (EMP) algorithm may be viewed as sum-product message passing over the entropy semiring, which has previously appeared in…

Machine Learning · Computer Science 2016-11-18 Velimir M. Ilic , Miomir S. Stankovic , Branimir T. Todorovic

The problem of determining the joint probability distributions for correlated random variables with pre-specified marginals is considered. When the joint distribution satisfying all the required conditions is not unique, the "most unbiased"…

Statistical Mechanics · Physics 2015-06-12 Hernán Larralde

When an expert operates a perilous dynamic system, ideal constraint information is tacitly contained in their demonstrated trajectories and controls. The likelihood of these demonstrations can be computed, given the system dynamics and task…

Systems and Control · Electrical Eng. & Systems 2021-02-26 David L. McPherson , Kaylene C. Stocking , S. Shankar Sastry

We propose an unsupervised method to extract keywords and keyphrases from texts based on a pre-trained language model (LM) and Shannon's information maximization. Specifically, our method extracts phrases having the highest conditional…

Computation and Language · Computer Science 2023-08-31 Alexander Tsvetkov , Alon Kipnis

Inferring a quantum system from incomplete information is a common problem in many aspects of quantum information science and applications, where the principle of maximum entropy (MaxEnt) plays an important role. The quantum state…

Quantum Physics · Physics 2022-07-26 Shi-Yao Hou , Zipeng Wu , Jinfeng Zeng , Ningping Cao , Chenfeng Cao , Youning Li , Bei Zeng

The asymptotic convergence of probability density function (pdf) and convergence of differential entropy are examined for the non-stationary processes that follow the maximum entropy principle (MaxEnt) and maximum entropy production…

Information Theory · Computer Science 2014-01-14 Alexander L. Fradkov , Dmitry S. Shalymov

Representation learning is currently a very hot topic in modern machine learning, mostly due to the great success of the deep learning methods. In particular low-dimensional representation which discriminates classes can not only enhance…

Machine Learning · Computer Science 2015-04-13 Wojciech Marian Czarnecki , Rafał Józefowicz , Jacek Tabor

Training Large Language Models (LLMs) with Group Relative Policy Optimization (GRPO) encounters a significant challenge: models often fail to produce accurate responses, particularly in small-scale architectures. This limitation not only…

Computation and Language · Computer Science 2025-10-10 Fu Chen , Peng Wang , Xiyin Li , Wen Li , Shichi Lei , Dongdong Xiang

IBM models are very important word alignment models in Machine Translation. Following the Maximum Likelihood Estimation principle to estimate their parameters, the models will easily overfit the training data when the data are sparse. While…

Computation and Language · Computer Science 2016-04-28 Vuong Van Bui , Cuong Anh Le

(Jaynes') Method of (Shannon-Kullback's) Relative Entropy Maximization (REM or MaxEnt) can be - at least in the discrete case - according to the Maximum Probability Theorem (MPT) viewed as an asymptotic instance of the Maximum Probability…

Data Analysis, Statistics and Probability · Physics 2012-08-27 M. Grendar, , M. Grendar

Prompt optimization is a practical and widely applicable alternative to fine tuning for improving large language model performance. Yet many existing methods evaluate candidate prompts by sampling full outputs, often coupled with self…

Computation and Language · Computer Science 2025-09-19 Chenzhuo Zhao , Ziqian Liu , Xinda Wang , Junting Lu , Chaoyi Ruan

A major challenge for community ecology is using spatio-temporal data to infer parameters of dynamical models without conducting laborious experiments. We present a novel framework from statistical physics -- Maximum Caliber -- to…

Populations and Evolution · Quantitative Biology 2025-11-07 Zachary Jackson , Mathew A. Leibold , Robert D. Holt , BingKan Xue

Modern language models (LMs) increasingly require two critical resources: computational resources and data resources. Data selection techniques can effectively reduce the amount of training data required for fine-tuning LMs. However, their…

Computation and Language · Computer Science 2026-02-20 Hongming Li , Yang Liu , Chao Huang

In this thesis we start by providing some detail regarding how we arrived at our present understanding of probabilities and how we manipulate them - the product and addition rules by Cox. We also discuss the modern view of entropy and how…

Data Analysis, Statistics and Probability · Physics 2009-01-21 Adom Giffin

We consider an extension of the conditional min- and max-entropies to infinite-dimensional separable Hilbert spaces. We show that these satisfy characterizing properties known from the finite-dimensional case, and retain…

Quantum Physics · Physics 2011-09-20 Fabian Furrer , Johan Aberg , Renato Renner

Multi-instance data, in which each object (bag) contains a collection of instances, are widespread in machine learning, computer vision, bioinformatics, signal processing, and social sciences. We present a maximum entropy (ME) framework for…

Machine Learning · Computer Science 2016-03-15 Behrouz Behmardi , Forrest Briggs , Xiaoli Z. Fern , Raviv Raich

Supervised topic models utilize document's side information for discovering predictive low dimensional representations of documents. Existing models apply the likelihood-based estimation. In this paper, we present a general framework of…

Machine Learning · Statistics 2013-04-09 Jun Zhu , Amr Ahmed , Eric P. Xing