Meta-SAC automatically tunes SAC's entropy temperature for better exploration.
problem Exploration-exploitation dilemma in reinforcement learning.
method Meta-SAC uses metagradient and a novel meta objective to automatically adjust SAC's entropy temperature.
result Meta-SAC outperforms SAC-v2 by 10% on the humanoid-v2 task.
Temperature scaling improves model uncertainty but not diversity in LLMs.
problem Improving the calibration and stochasticity of probabilistic models.
method Investigates theoretical properties of temperature scaling in classification and LLMs.
result Temperature scaling increases model uncertainty but not diversity in LLMs.
Enhances RL by controlling policy stochasticity through trajectory entropy constraints.
problem Non-stationary Q-value estimation and short-sighted entropy tuning in maximum entropy RL.
method Proposes TECRL framework with separate Q-functions for reward and entropy, enforcing a trajectory entropy constraint.
result DSAC-E algorithm achieves higher returns and better stability on OpenAI Gym benchmarks.
We translate the problem of calculating the entropy of a set of binary configurations/signals into a sequence of supervised classification tasks. Subsequently, one can use virtually any machine learning classification algorithm for computing entropy. This procedure can be used to compute entropy, and consequently the f…
The well known maximum-entropy principle due to Jaynes, which states that given mean parameters, the maximum entropy distribution matching them is in an exponential family, has been very popular in machine learning due to its "Occam's razor" interpretation. Unfortunately, calculating the potentials in the maximum-entro…
New statistical test for change-point detection using relative entropy.
problem Offline change-point detection using divergence metrics.
method Study of empirical relative entropy distributions, derivation of approximations, introduction of new Berry-Esseen bounds.
result Theoretical and practical validation of relative entropy for change-point detection.
Hyperfitting improves LLM generation quality by enhancing diversity, contrary to simple temperature scaling.
problem Improving open-ended generation quality of LLMs with minimal fine-tuning effort.
method Demonstrates that hyperfitting, a phenomenon where LLMs are fine-tuned to near-zero training loss, enhances generation quality and mitigates repetition.
result Hyperfitting is distinct from temperature scaling and involves a dynamic, context-dependent rank reordering mechanism in the final transformer block.
Study on non-archimedean μ-entropy for toric varieties, proving existence and uniqueness.
problem Exploring non-archimedean μ-entropy for toric varieties and its thermodynamical structure.
method Established a Rellich type compactness result for convex functions on simple polytope, proving existence and uniqueness of optimizer.
result Existence and uniqueness of optimizer for toric non-archimedean μ^λ-entropy for λ ≤ 0.
Unified framework for critical scaling of inverse temperature in self-attention.
problem Conflicting inverse-temperature laws for long-context self-attention.
method Counting gaps and defining an upper-tail accumulation scale.
result Critical inverse-temperature scale determined by gap-counting function.
New framework uses time series features for predicting streamflow in ungauged areas.
problem Predicting streamflow in areas without gauging stations.
method Developed regression-based streamflow regionalization using a wide range of time series features from large datasets.
result Certain time series features, like entropy and autocorrelation, are better predictors of streamflow than traditional catchment attributes.
Estimate relaxation times in nonextensive systems using gradient flow for Tsallis entropy maximization.
problem Estimating relaxation times in financial market dynamics.
method Developing a method using EGF for maximizing Tsallis entropy.
result Longer relaxation times for nonextensive systems compared to Shannon entropy.
Study on convergence of exponential probability measures with applications to maximum entropy models and SGLD.
problem Characterizing the limit of probability measures with exponential densities as temperature approaches zero.
method Quantitative bounds on Wasserstein distance using geometric measure theory tools.
result Established quantitative convergence results for norm-like potentials under invertibility conditions.
Information-theoretic quantities, such as entropy, are used to quantify the amount of information a given variable provides. Entropies can be used together to compute the mutual information, which quantifies the amount of information two variables share. However, accurately estimating these quantities from data is extr…
The coefficient of the logarithmic term in the entropy on even spheres is re-computed by the local technique of integrating the finite temperature energy density up to the horizon on static d--dimensional de Sitter space and thence finding the entropy by thermodynamics. Numeric evaluation yields the known answer i.e. (…
Coupled entropy corrects flaws in Tsallis entropy for complex systems.
problem Misinterpretation of generalized temperature and entropy.
method Derived from generalized Pareto and Student's t distributions.
result Provides balanced measure of uncertainty for complex systems.
Metric learning aims at learning a distance which is consistent with the semantic meaning of the samples. The problem is generally solved by learning an embedding for each sample such that the embeddings of samples of the same category are compact while the embeddings of samples of different categories are spread-out i…
Improved loss functions adapt to weight-space anisotropy, outperforming isotropic counterparts.
problem Adapting to the anisotropic nature of deep weight spaces for better performance.
method Refined local entropic loss functions restricted to a subset of weights, exploiting anisotropy.
result Partial local entropies outperform isotropic counterparts on image classification tasks.
NM-PPG optimizes adaptive feature acquisition in POMDPs for better predictions.
problem Optimizing adaptive feature acquisition in prediction problems with costly features.
method Non-myopic pathwise policy gradients (NM-PPG) with continuous relaxation and straight-through rollout.
result NM-PPG outperforms state-of-the-art AFA methods on synthetic and real-world datasets.
We introduce a novel Entropy-driven Monte Carlo (EdMC) strategy to efficiently sample solutions of random Constraint Satisfaction Problems (CSPs). First, we extend a recent result that, using a large-deviation analysis, shows that the geometry of the space of solutions of the Binary Perceptron Learning Problem (a proto…
Top-H decoding improves text generation by balancing creativity and coherence.
problem Balancing creativity and coherence in text generation.
method Entropy-constrained minimum divergence problem, NP-hard ECMM problem, top-H decoding.
result Top-H decoding outperforms min-p sampling by up to 25.63% on creative writing benchmarks.
MESMOC optimizes constrained multi-objective problems efficiently.
problem Constrained multi-objective optimization with expensive function evaluations.
method Max-value Entropy Search in the output space.
result MESMOC selects high-quality Pareto solutions efficiently.
Support vector machines (SVMs) naturally embody sparseness due to their use of hinge loss functions. However, SVMs can not directly estimate conditional class probabilities. In this paper we propose and study a family of coherence functions, which are convex and differentiable, as surrogates of the hinge function. The …
Study non-negative curvature Markov chains, proving entropy contraction.
problem Prove entropy contraction for Markov chains with non-negative curvature.
method Prove 1-step contraction in Wasserstein distance implies 1-step contraction in relative entropy.
result Prove MLSI with constant equal to minimal rate increment for mean-field zero-range process.
Study on utility maximization with Tsallis entropy in reinforcement learning.
problem Exploring utility maximization with Tsallis entropy in reinforcement learning.
method Introducing Tsallis entropy regularizer to induce exploration, investigating specific examples, characterizing well-posedness, designing reinforcement learning algorithm.
result Characterized well-posedness and provided semi-closed-form solutions for specific examples, found distinct optimal strategies.
Finding parameters that minimise a loss function is at the core of many machine learning methods. The Stochastic Gradient Descent algorithm is widely used and delivers state of the art results for many problems. Nonetheless, Stochastic Gradient Descent typically cannot find the global minimum, thus its empirical effect…
Deep networks have enabled reinforcement learning to scale to more complex and challenging domains, but these methods typically require large quantities of training data. An alternative is to use sample-efficient episodic control methods: neuro-inspired algorithms which use non-/semi-parametric models that predict valu…
New method resolves nonidentifiability in mixture models.
problem Nonidentifiability in marginal models of mixture models.
method Introducing an effective temperature to generalize the marginal likelihood.
result Maximization of the generalized likelihood leads to unique results.
SymCircuit learns PC structure via entropy-regularized RL, improving inference efficiency and accuracy.
problem Greedy algorithms in PC structure learning lead to suboptimal solutions.
method Entropy-regularized reinforcement learning to train a learned generative policy for PC structure inference.
result SymCircuit learns the optimal policy as a tempered Bayesian posterior, improving inference efficiency and accuracy.
This paper addresses two seemingly unrelated problems, (a) What is the entropy and energy accounting in the Maxwell Demon problem? and (b) How can the efficiency of markets be measured? Here we show, in a simple model for the Maxwell Demon, the entropy of the universe increases by an amount eta=0.839995520 in going fro…
New framework recovers reward and rationality parameters from game behavior.
problem Statistical ambiguity in identifying reward and rationality parameters in competitive games.
method Blind Inverse Game Theory (Blind-IGT) using entropy-regularized Quantal Response Equilibrium and Normalized Least Squares (NLS) estimator.
result Optimal convergence rate of O(N−1/2) for joint parameter recovery. Paper introduces DTAE to optimize RL algorithms, balancing exploration and exploitation.
problem Balancing exploration and exploitation in reinforcement learning.
method Soft policy optimization with entropy and dual-track advantage estimator (DTAE).
result DTAE accelerates RL algorithm convergence and improves performance.
New framework reveals thermodynamic principles for LLM training.
problem Understanding the training dynamics of large language models.
method Introducing Neural Thermodynamic Laws (NTL) under river-valley loss landscape assumptions.
result Key thermodynamic quantities and principles naturally emerge in LLM training.
Imaging techniques are essential tools for inquiring a number of properties from different materials. Liquid crystals are often investigated via optical and image processing methods. In spite of that, considerably less attention has been paid to the problem of extracting physical properties of liquid crystals directly …
Statistical mechanics explains income and wealth distribution in developed economies.
problem Understanding the distribution of income and wealth in developed economies.
method Derive the distribution from firm dynamics using maximum entropy and mixture aggregation.
result Derive the robust two-class structure of income and wealth distribution.
An algorithmic limit of compressed sensing or related variable-selection problems is analytically evaluated when a design matrix is given by an overcomplete random matrix. The replica method from statistical mechanics is employed to derive the result. The analysis is conducted through evaluation of the entropy, an expo…
Focal loss improves deep neural networks' accuracy and calibration.
problem Miscalibration in deep neural networks.
method Using focal loss and temperature scaling to improve model calibration.
result Focal loss leads to state-of-the-art calibrated models without sacrificing accuracy.
Introduce a thermodynamically informed, temperature-transferable MLCG framework for proteins.
problem Temperature transferability of MLCG models for proteins.
method Explicit decomposition of CG potential into energetic and entropic components.
result Reproduces temperature-dependent quantities like heat capacity.
We investigate unification of two systems of identical elements having different dimensions which may be of interest for both physics and economics. Characteristic parameters as well as explicit formulae for the temperature (in economics - capital turnover) and dimension of the united system are obtained as functions o…
Study local sensitivity of HDD and CDD temperature derivatives prices.
problem Understanding how temperature derivatives prices change with small temperature changes.
method Analyzes sensitivity of HDD and CDD futures and options prices to temperature perturbations using a CAR process.
result Identifies the order of the CAR process and its impact on temperature derivatives prices.
We introduce a temperature into the exponential function and replace the softmax output layer of neural nets by a high temperature generalization. Similarly, the logarithm in the log loss we use for training is replaced by a low temperature logarithm. By tuning the two temperatures we create loss functions that are non…
A new method selects optimal temperature for Bayesian Deep Learning.
problem Finding the optimal temperature for improving predictive performance in Bayesian Deep Learning.
method Data-driven approach to estimate temperature as a model parameter.
result Our method performs comparably to grid search but at a fraction of the cost.
Climate volatility reduces economic growth, especially in poorer countries.
problem Impact of climate volatility on economic growth.
method Exploiting data on 133 countries over 59 years, controlling for temperature changes.
result A 1 degree C increase in temperature volatility leads to a 0.3% decline in GDP growth.
New adaptive temperature selection improves parallel tempering efficiency.
problem Enhancing mixing in multi-modal distributions using parallel tempering.
method Adaptive temperature selection using policy gradient approach.
result Lower integrated autocorrelation times achieved compared to traditional methods.
Logit-GFN accelerates GFlowNets training by scaling logits based on temperature.
problem Training temperature-conditional GFlowNets is numerically challenging.
method Logit-GFN uses a learned function of temperature to scale policy logits.
result Logit-GFN greatly accelerates GFlowNets training and improves generalization and mode discovery.
This paper measures temperature in agent systems using volatility.
problem How to measure temperature in agent systems.
method Examined an agent system with two decision options in a news environment, established the measurement equation, and outlined the concept of temperature measurement.
result Illustrated a strategy for influencing average opinion in competing subsystems.
Personal income distribution in the USA has a well-defined two-class structure. The majority of population (97-99%) belongs to the lower class characterized by the exponential Boltzmann-Gibbs ("thermal") distribution, whereas the upper class (1-3% of population) has a Pareto power-law ("superthermal") distribution. By …
Attention temperature improves robustness of ICL in high-dimensional settings.
problem ICL robustness failure under distribution shift in high dimensions.
method Analyzed a Transformer with approximate softmax attention, derived a closed-form error expression, and showed optimal temperature minimizes error.
result Optimal attention temperature minimizes ICL generalization error under distribution shift.
Trends in terrestrial temperature variability are perhaps more relevant for species viability than trends in mean temperature. In this paper, we develop methodology for estimating such trends using multi-resolution climate data from polar orbiting weather satellites. We derive two novel algorithms for computation that …