Mechanistic interpretability work attempts to reverse engineer the learned algorithms present inside neural networks. One focus of this work has been to discover 'circuits' -- subgraphs of the full model that explain behaviour on specific tasks. But how do we measure the performance of such circuits? Prior work has attempted to measure circuit 'faithfulness' -- the degree to which the circuit replicates the performance of the full model. In this work, we survey many considerations for designing experiments that measure circuit faithfulness by ablating portions of the model's computation. Concerningly, we find existing methods are highly sensitive to seemingly insignificant changes in the ablation methodology. We conclude that existing circuit faithfulness scores reflect both the methodological choices of researchers as well as the actual components of the circuit - the task a circuit is required to perform depends on the ablation used to test it. The ultimate goal of mechanistic interpretability work is to understand neural networks, so we emphasize the need for more clarity in the precise claims being made about circuits. We open source a library at https://github.com/UFO-101/auto-circuit that includes highly efficient implementations of a wide range of ablation methodologies and circuit discovery algorithms.

This paper introduces a machine learning-aided fault detection and isolation method applied to the case study of quench identification at the European X-Ray Free-Electron Laser. The plant utilizes 800 superconducting radio-frequency cavities in order to accelerate electron bunches to high energies of up to 17.5 GeV. Various faulty events can disrupt the nominal functioning of the accelerator, including quenches that can lead to a loss of the superconductivity of the cavities and the interruption of their operation. In this context, our solution consists in analyzing signals reflecting the dynamics of the cavities in a two-stage approach. (I) Fault detection that uses analytical redundancy to process the data and generate a residual. The evaluation of the residual through the generalized likelihood ratio allows detecting the faulty behaviors. (II) Fault isolation which involves the distinction of the quenches from the other faults. To this end, we proceed with a data-driven model of the k-medoids algorithm that explores different similarity measures, namely, the Euclidean and the dynamic time warping. Finally, we evaluate the new method and compare it to the currently deployed quench detection system, the results show the improved performance achieved by our method.

Deep learning (DL) has emerged as a crucial tool in network anomaly detection (NAD) for cybersecurity. While DL models for anomaly detection excel at extracting features and learning patterns from data, they are vulnerable to data contamination -- the inadvertent inclusion of attack-related data in training sets presumed benign. This study evaluates the robustness of six unsupervised DL algorithms against data contamination using our proposed evaluation protocol. Results demonstrate significant performance degradation in state-of-the-art anomaly detection algorithms when exposed to contaminated data, highlighting the critical need for self-protection mechanisms in DL-based NAD models. To mitigate this vulnerability, we propose an enhanced auto-encoder with a constrained latent representation, allowing normal data to cluster more densely around a learnable center in the latent space. Our evaluation reveals that this approach exhibits improved resistance to data contamination compared to existing methods, offering a promising direction for more robust NAD systems.

In today's fast-paced world, accurately monitoring stress levels is crucial. Sensor-based stress monitoring systems often need large datasets for training effective models. However, individual-specific models are necessary for personalized and interactive scenarios. Traditional methods like Ecological Momentary Assessments (EMAs) assess stress but struggle with efficient data collection without burdening users. The challenge is to timely send EMAs, especially during stress, balancing monitoring efficiency and user convenience. This paper introduces a novel context-aware active reinforcement learning (RL) algorithm for enhanced stress detection using Photoplethysmography (PPG) data from smartwatches and contextual data from smartphones. Our approach dynamically selects optimal times for deploying EMAs, utilizing the user's immediate context to maximize label accuracy and minimize intrusiveness. Initially, the study was executed in an offline environment to refine the label collection process, aiming to increase accuracy while reducing user burden. Later, we integrated a real-time label collection mechanism, transitioning to an online methodology. This shift resulted in an 11% improvement in stress detection efficiency. Incorporating contextual data improved model accuracy by 4%. Personalization studies indicated a 10% enhancement in AUC-ROC scores, demonstrating better stress level differentiation. This research marks a significant move towards personalized, context-driven real-time stress monitoring methods.

Conventional anomaly detection techniques based on reconstruction via denoising diffusion model are widely used due to their ability to identify anomaly locations and shapes with high performance. However, there is a limitation in determining appropriate noise parameters that can degrade anomalies while preserving normal characteristics. Also, due to the volatility of the diffusion model, normal regions can fluctuate considerably during reconstruction, resulting in false detection. In this paper, we propose a method to detect anomalies by analysis of reconstruction trend depending on the degree of degradation, effectively solving the both problems of existing methods. The proposed method is validated on an open dataset for industrial anomaly detection, improving the performance of existing methods on a number of evaluation criteria. With the ease of combination with existing anomaly detection methods, it provides a tradeoff between computational cost and performance, allowing it high application potential in manufacturing industry.

This paper introduces Helios, the first extremely low-power, real-time, event-based hand gesture recognition system designed for all-day on smart eyewear. As augmented reality (AR) evolves, current smart glasses like the Meta Ray-Bans prioritize visual and wearable comfort at the expense of functionality. Existing human-machine interfaces (HMIs) in these devices, such as capacitive touch and voice controls, present limitations in ergonomics, privacy and power consumption. Helios addresses these challenges by leveraging natural hand interactions for a more intuitive and comfortable user experience. Our system utilizes a extremely low-power and compact 3mmx4mm/20mW event camera to perform natural hand-based gesture recognition for always-on smart eyewear. The camera's output is processed by a convolutional neural network (CNN) running on a NXP Nano UltraLite compute platform, consuming less than 350mW. Helios can recognize seven classes of gestures, including subtle microgestures like swipes and pinches, with 91% accuracy. We also demonstrate real-time performance across 20 users at a remarkably low latency of 60ms. Our user testing results align with the positive feedback we received during our recent successful demo at AWE-USA-2024.

Out-of-distribution (OOD) detection is crucial for model reliability, as it identifies samples from unknown classes and reduces errors due to unexpected inputs. Vision-Language Models (VLMs) such as CLIP are emerging as powerful tools for OOD detection by integrating multi-modal information. However, the practical application of such systems is challenged by manual prompt engineering, which demands domain expertise and is sensitive to linguistic nuances. In this paper, we introduce Label-driven Automated Prompt Tuning (LAPT), a novel approach to OOD detection that reduces the need for manual prompt engineering. We develop distribution-aware prompts with in-distribution (ID) class names and negative labels mined automatically. Training samples linked to these class labels are collected autonomously via image synthesis and retrieval methods, allowing for prompt learning without manual effort. We utilize a simple cross-entropy loss for prompt optimization, with cross-modal and cross-distribution mixing strategies to reduce image noise and explore the intermediate space between distributions, respectively. The LAPT framework operates autonomously, requiring only ID class names as input and eliminating the need for manual intervention. With extensive experiments, LAPT consistently outperforms manually crafted prompts, setting a new standard for OOD detection. Moreover, LAPT not only enhances the distinction between ID and OOD samples, but also improves the ID classification accuracy and strengthens the generalization robustness to covariate shifts, resulting in outstanding performance in challenging full-spectrum OOD detection tasks. Codes are available at \url{https://github.com/YBZh/LAPT}.

Federated Learning (FL) offers innovative solutions for privacy-preserving collaborative machine learning (ML). Despite its promising potential, FL is vulnerable to various attacks due to its distributed nature, affecting the entire life cycle of FL services. These threats can harm the model's utility or compromise participants' privacy, either directly or indirectly. In response, numerous defense frameworks have been proposed, demonstrating effectiveness in specific settings and scenarios. To provide a clear understanding of the current research landscape, this paper reviews the most representative and state-of-the-art threats and defense frameworks throughout the FL service life cycle. We start by identifying FL threats that harm utility and privacy, including those with potential or direct impacts. Then, we dive into the defense frameworks, analyze the relationship between threats and defenses, and compare the trade-offs among different defense strategies. Finally, we summarize current research bottlenecks and offer insights into future research directions to conclude this survey. We hope this survey sheds light on trustworthy FL research and contributes to the FL community.

Time-series data processing is an important component of many real-world applications, such as health monitoring, environmental monitoring, and digital agriculture. These applications collect distinct windows of sensor data (e.g., few seconds) and process them to assess the environment. Machine learning (ML) models are being employed in time-series applications due to their generalization abilities for classification. State-of-the-art time-series applications wait for entire sensor data window to become available before processing the data using ML algorithms, resulting in high sensor energy consumption. However, not all situations require processing full sensor window to make accurate inference. For instance, in activity recognition, sitting and standing activities can be inferred with partial windows. Using this insight, we propose to employ early exit classifiers with partial sensor windows to minimize energy consumption while maintaining accuracy. Specifically, we first utilize multiple early exits with successively increasing amount of data as they become available in a window. If early exits provide inference with high confidence, we return the label and enter low power mode for sensors. The proposed approach has potential to enable significant energy savings in time series applications. We utilize neural networks and random forest classifiers to evaluate our approach. Our evaluations with six datasets show that the proposed approach enables up to 50-60% energy savings on average without any impact on accuracy. The energy savings can enable time-series applications in remote locations with limited energy availability.

With the proliferation of the Internet of Things (IoT) and the rising interconnectedness of devices, network security faces significant challenges, especially from anomalous activities. While traditional machine learning-based intrusion detection systems (ML-IDS) effectively employ supervised learning methods, they possess limitations such as the requirement for labeled data and challenges with high dimensionality. Recent unsupervised ML-IDS approaches such as AutoEncoders and Generative Adversarial Networks (GAN) offer alternative solutions but pose challenges in deployment onto resource-constrained IoT devices and in interpretability. To address these concerns, this paper proposes a novel federated unsupervised anomaly detection framework, FedPCA, that leverages Principal Component Analysis (PCA) and the Alternating Directions Method Multipliers (ADMM) to learn common representations of distributed non-i.i.d. datasets. Building on the FedPCA framework, we propose two algorithms, FEDPE in Euclidean space and FEDPG on Grassmann manifolds. Our approach enables real-time threat detection and mitigation at the device level, enhancing network resilience while ensuring privacy. Moreover, the proposed algorithms are accompanied by theoretical convergence rates even under a subsampling scheme, a novel result. Experimental results on the UNSW-NB15 and TON-IoT datasets show that our proposed methods offer performance in anomaly detection comparable to nonlinear baselines, while providing significant improvements in communication and memory efficiency, underscoring their potential for securing IoT networks.