arXiv research

Bayesian neural networks use temperature adjustments to improve predictive performance.

New adaptive temperature selection improves parallel tempering efficiency.

Attention temperature improves robustness of ICL in high-dimensional settings.

Meta-SAC automatically tunes SAC's entropy temperature for better exploration.

AdaAnn optimizes annealing for efficient probability density approximation.

A model predicts how hyperparameters affect pruning performance.

Minimizing non-convex and high-dimensional objective functions is challenging, especially when training modern deep neural networks. In this paper, a novel approach is proposed which divides the training process into two consecutive phases to obtain better generalization performance: Bayesian sampling and stochastic op…

Enhances RL by controlling policy stochasticity through trajectory entropy constraints.

Study evaluates different mathematical models for three case studies using statistical fitting.

We present a novel Metropolis-Hastings method for large datasets that uses small expected-size minibatches of data. Previous work on reducing the cost of Metropolis-Hastings tests yield variable data consumed per sample, with only constant factor reductions versus using the full dataset for each sample. Here we present…

Temperature scaling improves model uncertainty but not diversity in LLMs.

Introduce a thermodynamically informed, temperature-transferable MLCG framework for proteins.

Developing a climate-aware pricing framework for XL reinsurance and CAT bonds under non-stationary catastrophe risk.

Study local sensitivity of HDD and CDD 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.

Climate volatility reduces economic growth, especially in poorer countries.

Logit-GFN accelerates GFlowNets training by scaling logits based on temperature.

This paper measures temperature in agent systems using volatility.

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 …

The paper analyzes extreme temperature forecasting using machine learning models.

The paper analyzes the InfoNCE loss under different temperature schedules using Langevin dynamics.

This paper investigates uncertainty calibration in multimodal large language models.

The Gibbs algorithm's generalization error is bounded, improving with prior volume in low temperatures.

Optimizes contrastive learning with individualized temperatures for better performance on imbalanced datasets.

New method uses RBM flows to find critical temperatures in Ising models.

The paper prices weather contracts using a complex temperature model.

Algorithm learns Sherrington-Kirkpatrick model parameters at low temperatures.

Stochastic Gradient Descent introduces noise in training, affecting model decision boundaries.

Develops a new stochastic volatility model for temperature derivatives.

Power-SMC reduces inference latency for training-free LLM reasoning.

Study shows sample complexity for logistic regression with normal covariates.

The minute fluctuations of of S&P 500 and NASDAQ 100 indices display Boltzmann statistics over a wide range of positive as well as negative returns, thus allowing us to define a {\em market temperature} for either sign. With increasing time the sharp Boltzmann peak broadens into a Gaussian whose volatility $σ$ measure…

In the spirit of behavioral finance, we study the process of opinion formation among investors using a variant of the 2D Voter Model with a tunable social temperature. Further, a feedback acting on the temperature is introduced, such that social temperature reacts to market imbalances and thus becomes time dependent. I…

A spin model relating physical to financial variables is presented. This work is the first to introduce the concept of negative absolute temperature into stock market dynamics by establishing a rigorous formal analogy between physical and financial variables. Based on this model, an algorithm evaluating negative temper…

Machine learning and deep learning infer surface/groundwater exchange from temperature data.

Researchers derive the relation between temperature and volatility in ideal agent systems.

New method estimates Nishimori temperature for node classification in weighted graphs.

Cold posteriors improve Bayesian neural networks by reducing overestimation of aleatoric uncertainty.

Temperature scaling fails for distributions with class overlaps, while Mixup improves calibration.

The uncertainty estimation is critical in real-world decision making applications, especially when distributional shift between the training and test data are prevalent. Many calibration methods in the literature have been proposed to improve the predictive uncertainty of DNNs which are generally not well-calibrated. H…

Understanding local currents in the North Atlantic region of the ocean is a key part of modelling heat transfer and global climate patterns. Satellites provide a surface signature of the temperature of the ocean with a high horizontal resolution while in situ autonomous probes supply high vertical resolution, but horiz…

We present a model of financial markets originally proposed for a turbulent flow, as a dynamic basis of its intermittent behavior. Time evolution of the price change is assumed to be described by Brownian motion in a power-law potential, where the `temperature' fluctuates slowly. The model generally yields a fat-tailed…

Improved PDMP samplers for multi-modal distributions using tempering.

LiST improves neural network robustness and calibration without manual tuning.

Unified framework for critical scaling of inverse temperature in self-attention.

Model uncertainty obtained by variational Bayesian inference with Monte Carlo dropout is prone to miscalibration. The uncertainty does not represent the model error well. In this paper, temperature scaling is extended to dropout variational inference to calibrate model uncertainty. Expected uncertainty calibration erro…

Polynomial mixing times for simulated tempering in mixture sampling problems.