A key challenge in training Large Language Models (LLMs) is properly aligning them with human preferences. Reinforcement Learning with Human Feedback (RLHF) uses pairwise comparisons from human annotators to train reward functions and has emerged as a popular alignment method. However, input datasets in RLHF are not necessarily balanced in the types of questions and answers that are included. Therefore, we want RLHF algorithms to perform well even when the set of alternatives is not uniformly distributed. Drawing on insights from social choice theory, we introduce robustness to approximate clones, a desirable property of RLHF algorithms which requires that adding near-duplicate alternatives does not significantly change the learned reward function. We first demonstrate that the standard RLHF algorithm based on regularized maximum likelihood estimation (MLE) fails to satisfy this property. We then propose the weighted MLE, a new RLHF algorithm that modifies the standard regularized MLE by weighting alternatives based on their similarity to other alternatives. This new algorithm guarantees robustness to approximate clones while preserving desirable theoretical properties.

The development of neural sequence models, such as Transformers [Vaswani et al., 2017], along with the improvements in large-scale self-supervised learning, has opened the door to universal language understanding and generation. This achievement is largely motivated by pre-training: we separate common components from many neural network-based systems, and then train them on huge amounts of unlabeled data using self-supervision. These pre-trained models serve as foundation models that can be easily adapted to different tasks via fine-tuning or prompting. As a result, the paradigm of NLP has been enormously changed. In many cases, large-scale supervised learning for specific tasks is no longer required, and instead, we only need to adapt pre-trained foundation models.

We propose that inherent stochasticity enables synaptic plasticity to carry out probabilistic inference by sampling from a posterior distribution of synaptic parameters. This view provides a viable alternative to existing models that propose convergence of synaptic weights to maximum likelihood parameters. It explains how priors on weight distributions and connection probabilities can be merged optimally with learned experience. In simulations we show that our model for synaptic plasticity allows spiking neural networks to compensate continuously for unforeseen disturbances. Furthermore it provides a normative mathematical framework to better understand the permanent variability and rewiring observed in brain networks.

Recent work proposed the computation of so-called PI-explanations of Naive Bayes Classifiers (NBCs). PI-explanations are subset-minimal sets of feature-value pairs that are sufficient for the prediction, and have been computed with state-of-the-art exact algorithms that are worst-case exponential in time and space. In contrast, we show that the computation of one PI-explanation for an NBC can be achieved in log-linear time, and that the same result also applies to the more general class of linear classifiers. Furthermore, we show that the enumeration of PI-explanations can be obtained with polynomial delay. Experimental results demonstrate the performance gains of the new algorithms when compared with earlier work.

Inputs to machine learning models can have associated noise or uncertainties, but they are often ignored and not modelled. It is unknown if Bayesian Neural Networks and their approximations are able to consider uncertainty in their inputs. In this paper we build a two input Bayesian Neural Network (mean and standard deviation) and evaluate its capabilities for input uncertainty estimation across different methods like Ensembles, MC-Dropout, and Flipout. Our results indicate that only some uncertainty estimation methods for approximate Bayesian NNs can model input uncertainty, in particular Ensembles and Flipout.

This paper presents a new hybrid model for predicting German electricity prices. The algorithm is based on combining Gaussian Process Regression (GPR) and Support Vector Regression (SVR). While GPR is a competent model for learning the stochastic pattern within the data and interpolation, its performance for out-of-sample data is not very promising. By choosing a suitable data-dependent covariance function, we can enhance the performance of GPR for the tested German hourly power prices. However, since the out-of-sample prediction depends on the training data, the prediction is vulnerable to noise and outliers. To overcome this issue, a separate prediction is made using SVR, which applies margin-based optimization, having an advantage in dealing with non-linear processes and outliers, since only certain necessary points (support vectors) in the training data are responsible for regression. Both individual predictions are later combined using the performance-based weight assignment method. A test on historic German power prices shows that this approach outperforms its chosen benchmarks such as the autoregressive exogenous model, the naive approach, as well as the long short-term memory approach of prediction.

We study the differences arising from merging predictors in the causal and anticausal directions using the same data. In particular we study the asymmetries that arise in a simple model where we merge the predictors using one binary variable as target and two continuous variables as predictors. We use Causal Maximum Entropy (CMAXENT) as inductive bias to merge the predictors, however, we expect similar differences to hold also when we use other merging methods that take into account asymmetries between cause and effect. We show that if we observe all bivariate distributions, the CMAXENT solution reduces to a logistic regression in the causal direction and Linear Discriminant Analysis (LDA) in the anticausal direction. Furthermore, we study how the decision boundaries of these two solutions differ whenever we observe only some of the bivariate distributions implications for Out-Of-Variable (OOV) generalisation.

Across the scientific realm, we find ourselves subtracting or dividing stochastic signals. For instance, consider a stochastic realization, $x$, generated from the addition or multiplication of two stochastic signals $a$ and $b$, namely $x=a+b$ or $x = ab$. For the $x=a+b$ example, $a$ can be fluorescence background and $b$ the signal of interest whose statistics are to be learned from the measured $x$. Similarly, when writing $x=ab$, $a$ can be thought of as the illumination intensity and $b$ the density of fluorescent molecules of interest. Yet dividing or subtracting stochastic signals amplifies noise, and we ask instead whether, using the statistics of $a$ and the measurement of $x$ as input, we can recover the statistics of $b$. Here, we show how normalizing flows can generate an approximation of the probability distribution over $b$, thereby avoiding subtraction or division altogether. This method is implemented in our software package, NFdeconvolve, available on GitHub with a tutorial linked in the main text.

We observe that BatchBALD, a popular acquisition function for batch Bayesian active learning for classification, can conflate epistemic and aleatoric uncertainty, leading to suboptimal performance. Motivated by this observation, we propose to focus on the predictive probabilities, which only exhibit epistemic uncertainty. The result is an acquisition function that not only performs better, but is also faster to evaluate, allowing for larger batches than before.

Human Activity Recognition (HAR) has gained significant importance with the growing use of sensor-equipped devices and large datasets. This paper evaluates the performance of three categories of models : classical machine learning, deep learning architectures, and Restricted Boltzmann Machines (RBMs) using five key benchmark datasets of HAR (UCI-HAR, OPPORTUNITY, PAMAP2, WISDM, and Berkeley MHAD). We assess various models, including Decision Trees, Random Forests, Convolutional Neural Networks (CNN), and Deep Belief Networks (DBNs), using metrics such as accuracy, precision, recall, and F1-score for a comprehensive comparison. The results show that CNN models offer superior performance across all datasets, especially on the Berkeley MHAD. Classical models like Random Forest do well on smaller datasets but face challenges with larger, more complex data. RBM-based models also show notable potential, particularly for feature learning. This paper offers a detailed comparison to help researchers choose the most suitable model for HAR tasks.