In a recent paper, Erasmus et al. (2021) defend the idea that the ambiguity of the term "explanation" in explainable AI (XAI) can be solved by adopting any of four different extant accounts of explanation in the philosophy of science: the Deductive Nomological, Inductive Statistical, Causal Mechanical, and New Mechanist models. In this chapter, I show that the authors' claim that these accounts can be applied to deep neural networks as they would to any natural phenomenon is mistaken. I also provide a more general argument as to why the notion of explainability as it is currently used in the XAI literature bears little resemblance to the traditional concept of scientific explanation. It would be more fruitful to use the label "understandable AI" to avoid the confusion that surrounds the goal and purposes of XAI. In the second half of the chapter, I argue for a pragmatic conception of understanding that is better suited to play the central role attributed to explanation in XAI. Following Kuorikoski & Ylikoski (2015), the conditions of satisfaction for understanding an ML system are fleshed out in terms of an agent's success in using the system, in drawing correct inferences from it.

Spiking Neural Networks (SNN). SNNs are based on a more biologically inspired approach than usual artificial neural networks. Such models are characterized by complex dynamics between neurons and spikes. These are very sensitive to the hyperparameters, making their optimization challenging. To tackle hyperparameter optimization of SNNs, we initially extended the signal loss issue of SNNs to what we call silent networks. These networks fail to emit enough spikes at their outputs due to mistuned hyperparameters or architecture. Generally, search spaces are heavily restrained, sometimes even discretized, to prevent the sampling of such networks. By defining an early stopping criterion detecting silent networks and by designing specific constraints, we were able to instantiate larger and more flexible search spaces. We applied a constrained Bayesian optimization technique, which was asynchronously parallelized, as the evaluation time of a SNN is highly stochastic. Large-scale experiments were carried-out on a multi-GPU Petascale architecture. By leveraging silent networks, results show an acceleration of the search, while maintaining good performances of both the optimization algorithm and the best solution obtained. We were able to apply our methodology to two popular training algorithms, known as spike timing dependent plasticity and surrogate gradient. Early detection allowed us to prevent worthless and costly computation, directing the search toward promising hyperparameter combinations. Our methodology could be applied to multi-objective problems, where the spiking activity is often minimized to reduce the energy consumption. In this scenario, it becomes essential to find the delicate frontier between low-spiking and silent networks. Finally, our approach may have implications for neural architecture search, particularly in defining suitable spiking architectures.

The problem of search relevance in the E-commerce domain is a challenging one since it involves understanding the intent of a user's short nuanced query and matching it with the appropriate products in the catalog. This problem has traditionally been addressed using language models (LMs) and graph neural networks (GNNs) to capture semantic and inter-product behavior signals, respectively. However, the rapid development of new architectures has created a gap between research and the practical adoption of these techniques. Evaluating the generalizability of these models for deployment requires extensive experimentation on complex, real-world datasets, which can be non-trivial and expensive. Furthermore, such models often operate on latent space representations that are incomprehensible to humans, making it difficult to evaluate and compare the effectiveness of different models. This lack of interpretability hinders the development and adoption of new techniques in the field. To bridge this gap, we propose Plug and Play Graph LAnguage Model (PP-GLAM), an explainable ensemble of plug and play models. Our approach uses a modular framework with uniform data processing pipelines. It employs additive explanation metrics to independently decide whether to include (i) language model candidates, (ii) GNN model candidates, and (iii) inter-product behavioral signals. For the task of search relevance, we show that PP-GLAM outperforms several state-of-the-art baselines as well as a proprietary model on real-world multilingual, multi-regional e-commerce datasets. To promote better model comprehensibility and adoption, we also provide an analysis of the explainability and computational complexity of our model. We also provide the public codebase and provide a deployment strategy for practical implementation.

When large language models are trained on private data, it can be a significant privacy risk for them to memorize and regurgitate sensitive information. In this work, we propose a new practical data extraction attack that we call "neural phishing". This attack enables an adversary to target and extract sensitive or personally identifiable information (PII), e.g., credit card numbers, from a model trained on user data with upwards of 10% attack success rates, at times, as high as 50%. Our attack assumes only that an adversary can insert as few as 10s of benign-appearing sentences into the training dataset using only vague priors on the structure of the user data. Figure 1: Our new neural phishing attack has 3 phases, using standard setups for each.

Autonomous self-improving robots that interact and improve with experience are key to the real-world deployment of robotic systems. In this paper, we propose an online learning method, SELFI, that leverages online robot experience to rapidly fine-tune pre-trained control policies efficiently. SELFI applies online model-free reinforcement learning on top of offline model-based learning to bring out the best parts of both learning paradigms. Specifically, SELFI stabilizes the online learning process by incorporating the same model-based learning objective from offline pre-training into the Q-values learned with online model-free reinforcement learning. We evaluate SELFI in multiple real-world environments and report improvements in terms of collision avoidance, as well as more socially compliant behavior, measured by a human user study. SELFI enables us to quickly learn useful robotic behaviors with less human interventions such as pre-emptive behavior for the pedestrians, collision avoidance for small and transparent objects, and avoiding travel on uneven floor surfaces. We provide supplementary videos to demonstrate the performance of our fine-tuned policy on our project page.

We introduce a multi-step reasoning framework using prompt-based LLMs to examine the relationship between social media language patterns and trends in national health outcomes. Grounded in fuzzy-trace theory, which emphasizes the importance of gists of causal coherence in effective health communication, we introduce Role-Based Incremental Coaching (RBIC), a prompt-based LLM framework, to identify gists at-scale. Using RBIC, we systematically extract gists from subreddit discussions opposing COVID-19 health measures (Study 1). We then track how these gists evolve across key events (Study 2) and assess their influence on online engagement (Study 3). Finally, we investigate how the volume of gists is associated with national health trends like vaccine uptake and hospitalizations (Study 4). Our work is the first to empirically link social media linguistic patterns to real-world public health trends, highlighting the potential of prompt-based LLMs in identifying critical online discussion patterns that can form the basis of public health communication strategies.

In this study, we leverage SCADA data from diverse wind turbines to predict power output, employing advanced time series methods, specifically Functional Neural Networks (FNN) and Long Short-Term Memory (LSTM) networks. A key innovation lies in the ensemble of FNN and LSTM models, capitalizing on their collective learning. This ensemble approach outperforms individual models, ensuring stable and accurate power output predictions. Additionally, machine learning techniques are applied to detect wind turbine performance deterioration, enabling proactive maintenance strategies and health assessment. Crucially, our analysis reveals the uniqueness of each wind turbine, necessitating tailored models for optimal predictions. These insight underscores the importance of providing automatized customization for different turbines to keep human modeling effort low. Importantly, the methodologies developed in this analysis are not limited to wind turbines; they can be extended to predict and optimize performance in various machinery, highlighting the versatility and applicability of our research across diverse industrial contexts.

The diversity across outputs generated by large language models shapes the perception of their quality and utility. Prompt leaks, templated answer structure, and canned responses across different interactions are readily noticed by people, but there is no standard score to measure this aspect of model behavior. In this work we empirically investigate diversity scores on English texts. We find that computationally efficient compression algorithms capture information similar to what is measured by slow to compute $n$-gram overlap homogeneity scores. Further, a combination of measures -- compression ratios, self-repetition of long $n$-grams and Self-BLEU and BERTScore -- are sufficient to report, as they have low mutual correlation with each other. The applicability of scores extends beyond analysis of generative models; for example, we highlight applications on instruction-tuning datasets and human-produced texts. We release a diversity score package to facilitate research and invite consistency across reports.

Communication overhead is a significant bottleneck in federated learning (FL), which has been exaggerated with the increasing size of AI models. In this paper, we propose FedRDMA, a communication-efficient cross-silo FL system that integrates RDMA into the FL communication protocol. To overcome the limitations of RDMA in wide-area networks (WANs), FedRDMA divides the updated model into chunks and designs a series of optimization techniques to improve the efficiency and robustness of RDMA-based communication. We implement FedRDMA atop the industrial federated learning framework and evaluate it on a real-world cross-silo FL scenario. The experimental results show that \sys can achieve up to 3.8$\times$ speedup in communication efficiency compared to traditional TCP/IP-based FL systems.

Flaky tests can pass or fail non-deterministically, without alterations to a software system. Such tests are frequently encountered by developers and hinder the credibility of test suites. State-of-the-art research incorporates machine learning solutions into flaky test detection and achieves reasonably good accuracy. Moreover, the majority of automated flaky test repair solutions are designed for specific types of flaky tests. This research work proposes a novel categorization framework, called FlaKat, which uses machine-learning classifiers for fast and accurate prediction of the category of a given flaky test that reflects its root cause. Sampling techniques are applied to address the imbalance between flaky test categories in the International Dataset of Flaky Test (IDoFT). A new evaluation metric, called Flakiness Detection Capacity (FDC), is proposed for measuring the accuracy of classifiers from the perspective of information theory and provides proof for its effectiveness. The final FDC results are also in agreement with F1 score regarding which classifier yields the best flakiness classification.