Search

In this paper, we propose the $\textit{geometric invariance hypothesis (GIH)}$, which argues that when training a neural network, the input space curvature remains invariant under transformation in certain directions determined by its architecture. Starting with a simple non-linear binary classification problem residing on a plane in a high dimensional space, we observe that while an MLP can solve this problem regardless of the orientation of the plane, this is not the case for a ResNet. Motivated by this example, we define two maps that provide a compact $\textit{architecture-dependent}$ summary of the input space geometry of a neural network and its evolution during training, which we dub the $\textbf{average geometry}$ and $\textbf{average geometry evolution}$, respectively. By investigating average geometry evolution at initialization, we discover that the geometry of a neural network evolves according to the projection of data covariance onto average geometry. As a result, in cases where the average geometry is low-rank (such as in a ResNet), the geometry only changes in a subset of the input space. This causes an architecture-dependent invariance property in input-space curvature, which we dub GIH. Finally, we present extensive experimental results to observe the consequences of GIH and how it relates to generalization in neural networks.

Electric vehicle (EV) batteries are often underutilized. Vehicle-to-grid (V2G) services can tap into this unused potential, but increased battery usage may lead to more degradation and shorter battery life. This paper substantiates the advantages of providing load-shifting V2G services when the battery is aging, primarily due to calendar aging mechanisms (active degradation mechanisms while the battery is not used). After parameterizing a physics-based digital-twin for three different dominant degradation patterns within the same chemistry (NMC), we introduce a novel metric for evaluating the benefit and associated harm of V2G services: \textit{throughput gained versus days lost (TvD)} and show its strong relationship to the ratio of loss of lithium inventory (LLI) due to calendar aging to the total LLI ($\text{LLI}_\text{Cal}/\text{LLI}$). Our results that focus systematically on degradation mechanisms via lifetime simulation of digital-twins significantly expand prior work that was primarily concentrating on quantifying and reducing the degradation of specific cells by probing their usage and charging patterns. Examining various cell chemistries and conditions enables us to take a broader view and determine whether a particular battery pack is appropriate for load-shifting (V2G) services. Our research demonstrates that the decision "to V2G or not to V2G" can be made by merely estimating the portion of capacity deterioration caused by calendar aging. Specifically, TvD is primarily influenced by the chemistry of cells and the environmental temperature where the car is parked, while the usage intensity and charging patterns of EVs play a lesser role.

The battery state of health (SOH) based on capacity fade and resistance increase is not sufficient for predicting Remaining Useful life (RUL). The electrochemical community blames the path-dependency of the battery degradation mechanisms for our inability to forecast the degradation. The control community knows that the path-dependency is addressed by full state estimation. We show that even the electrode-specific SOH (eSOH) estimation is not enough to fully define the degradation states by simulating infinite possible degradation trajectories and remaining useful lives (RUL) from a unique eSOH. We finally define the deepSOH states that capture the individual contributions of all the common degradation mechanisms, namely, SEI, plating, and mechanical fracture to the loss of lithium inventory. We show that the addition of cell expansion measurement may allow us to estimate the deepSOH and predict the remaining useful life., Comment: 6 pages, 3 figures

Background: Synthetic data has been proposed as a solution for sharing anonymized versions of sensitive biomedical datasets. Ideally, synthetic data should preserve the structure and statistical properties of the original data, while protecting the privacy of the individual subjects. Differential privacy (DP) is currently considered the gold standard approach for balancing this trade-off. Objectives: To investigate the reliability of group differences identified by independent sample tests on DP-synthetic data. The evaluation is conducted in terms of the tests' Type I and Type II errors. The former quantifies the tests' validity i.e. whether the probability of false discoveries is indeed below the significance level, and the latter indicates the tests' power in making real discoveries. Methods: We evaluate the Mann-Whitney U test, Student's t-test, chi-squared test and median test on DP-synthetic data. The private synthetic datasets are generated from real-world data, including a prostate cancer dataset (n=500) and a cardiovascular dataset (n=70 000), as well as on bivariate and multivariate simulated data. Five different DP-synthetic data generation methods are evaluated, including two basic DP histogram release methods and MWEM, Private-PGM, and DP GAN algorithms. Conclusion: A large portion of the evaluation results expressed dramatically inflated Type I errors, especially at privacy budget levels of $\epsilon\leq 1$. This result calls for caution when releasing and analyzing DP-synthetic data: low p-values may be obtained in statistical tests simply as a byproduct of the noise added to protect privacy. A DP smoothed histogram-based synthetic data generation method was shown to produce valid Type I error for all privacy levels tested but required a large original dataset size and a modest privacy budget ($\epsilon\geq 5$) in order to have reasonable Type II error.

Project managers play a crucial role in the success of projects. The selection of an appropriate project manager is a primary concern for senior managers in firms. Typically, this process involves candidate interviews and assessments of their abilities. There are various criteria for selecting a project manager, and the importance of each criterion depends on the project type, its conditions, and the risks associated with their absence in the chosen candidate. Often, senior managers in engineering companies lack awareness of the significance of these criteria and the potential risks linked to their absence. This research aims to identify these risks in selecting project managers for civil engineering projects, utilizing a combined ANP-FMEA approach. Through a comprehensive literature review, five risk categories have been identified: individual skills, power-related issues, knowledge and expertise, experience, and personality traits. Subsequently, these risks, along with their respective sub-criteria and internal relationships, were analysed using the combined ANP-FMEA technique. The results highlighted that the lack of political influence, absence of construction experience, and deficiency in project management expertise represent the most substantial risks in selecting a project manager. Moreover, upon comparison with the traditional FMEA approach, this study demonstrates the superior ability of the ANP-FMEA model in differentiating risks and pinpointing factors with elevated risk levels., Comment: 16 pages

This paper proposes an analytical framework describing how initial capacity and resistance variability in parallel-connected battery cells may inflict additional variability or reduce variability while the cells age. We derive closed-form equations for current and SOC imbalance dynamics within a charge or discharge cycle. These dynamics are represented by a first-order equivalent circuit model and validated against experimental data. To demonstrate how current and SOC imbalance leads to cell degradation, we developed a successive update scheme in which the inter-cycle imbalance dynamics update the intra-cycle degradation dynamics, and vice versa. Using this framework, we demonstrate that current imbalance can cause convergent degradation trajectories, consistent with previous reports. However, we also demonstrate that different degradation assumptions, such as those associated with SOC imbalance, may cause divergent degradation. We finally highlight the role of different cell chemistries, including different OCV function nonlinearities, on system behavior, and derive analytical bounds on the SOC imbalance using Lyapunov analysis., Comment: 22 pages, 14 figures, submitted to the Journal of Dynamic Systems, Measurement and Control

Very large numbers of M-pox cases have, since the start of May 2022, been reported in non-endemic countries leading many to fear that the M-pox Outbreak would rapidly transition into another pandemic, while the COVID-19 pandemic ravages on. Given the similarities of M-pox with COVID-19, we chose to test the performance of COVID-19 models trained on South African twitter data on a hand-labelled M-pox dataset before and after fine-tuning. More than 20k M-pox-related tweets from South Africa were hand-labelled as being either positive, negative or neutral. After fine-tuning these COVID-19 models on the M-pox dataset, the F1-scores increased by more than 8% falling just short of 70%, but still outperforming state-of-the-art models and well-known classification algorithms. An LDA-based topic modelling procedure was used to compare the miss-classified M-pox tweets of the original COVID-19 RoBERTa model with its fine-tuned version, and from this analysis, we were able to draw conclusions on how to build more sophisticated models.

Very few social media studies have been done on South African user-generated content during the COVID-19 pandemic and even fewer using hand-labelling over automated methods. Vaccination is a major tool in the fight against the pandemic, but vaccine hesitancy jeopardizes any public health effort. In this study, sentiment analysis on South African tweets related to vaccine hesitancy was performed, with the aim of training AI-mediated classification models and assessing their reliability in categorizing UGC. A dataset of 30000 tweets from South Africa were extracted and hand-labelled into one of three sentiment classes: positive, negative, neutral. The machine learning models used were LSTM, bi-LSTM, SVM, BERT-base-cased and the RoBERTa-base models, whereby their hyperparameters were carefully chosen and tuned using the WandB platform. We used two different approaches when we pre-processed our data for comparison: one was semantics-based, while the other was corpus-based. The pre-processing of the tweets in our dataset was performed using both methods, respectively. All models were found to have low F1-scores within a range of 45$\%$-55$\%$, except for BERT and RoBERTa which both achieved significantly better measures with overall F1-scores of 60$\%$ and 61$\%$, respectively. Topic modelling using an LDA was performed on the miss-classified tweets of the RoBERTa model to gain insight on how to further improve model accuracy.

Let $G=(V, E)$ be a simple graph with vertex set $V$ and edge set $E$. The Sombor index of the graph $G$ is a degree-based topological index, defined as $$SO(G)=\sum_{uv \in E}\sqrt{d(u)^2+d(v)^2},$$ in which $d(x)$ is the degree of the vertex $x \in V$ for $x=u, v$. In this paper, we characterize the extremal trees with a given degree sequence that maximizes the Sombor index.

While deep learning in the form of recurrent neural networks (RNNs) has caused a significant improvement in neural language modeling, the fact that they are extremely prone to overfitting is still a mainly unresolved issue. In this paper we propose a regularization method based on generative adversarial networks (GANs) and adversarial training (AT), that can prevent overfitting in neural language models. Unlike common adversarial training methods such as the fast gradient sign method (FGSM) that require a second back-propagation through time, and therefore effectively require at least twice the amount of time for regular training, the overhead of our method does not exceed more than 20% of the training of the baselines., Comment: An extended abstract selected for poster presentation at the Eastern European Machine Learning Summer School 2019