Afterward, we offer a solution that is capable of 1) making our AI systems robust against deepfakes during development and deployment phases; 2) detecting video, image, audio, and textual deepfakes; 3) identifying deepfakes that bypass detection (deepfake hunting); 4) leveraging available intelligence for timely identification of deepfake campaigns launched by state-sponsored hacking teams; 5) conducting in-depth forensic analysis of identified deepfake payloads. Our proposed solution can be used as a technical guide for developing detection, deterrence, and forensics investigation solutions for deepfakes. Our solution would address important elements of Canada's National Cyber Security Action Plan (2019-2024) in increasing the trustworthiness of our critical services [5]. Following actions can be taken based on this research findings: Raising public awareness about risks of deepfakes: increasing the understanding of deepfake threats and empowering Canadian public to do their part in keeping our society and critical services safe from deepfake-based attacks is the most important and effective step in reducing risk of deepfakes. Cybersecurity should always be considered as a shared responsibility. While this paper is focused on development of technical solutions for early detection and deterrence of deepfakes, the effectiveness of our solutions (or any technical solution in cybersecurity) are limited without regular and systemic public awareness campaigns. Supporting development of public training programs in this domain should be considered as a top priority. Developing AI robustness monitoring solutions: there is a growing trend in using AI to detect deepfakes. However, more recently, adversaries made attempts to create adversarial deepfake payloads that are capable of deceiving humans while bypassing AI-based detection systems!

Convolutional neural networks (CNNs) are an incredibly useful tool for analysing pictures, and in this article, we attempt to use one to identify breed given an image of a dog. On top of this, we also aim to input pictures of humans into the model and output the breed the human looks most similar to. To input into the model, we were provided with over 8,000 dog images each accompanied with the corresponding breed -- a total of 133 breeds over the whole dataset. To undertake this task, it was important to understand the theory behind CNNs and how they work, with particular application to how they work for image classification. The first consideration is how images can be represented for input to a CNN.

In astronomical surveys, such as the Zwicky Transient Facility, supernovae (SNe) are relatively uncommon objects compared to other classes of variable events. Along with this scarcity, the processing of multi-band light-curves is a challenging task due to the highly irregular cadence, long time gaps, missing-values, few observations, etc. These issues are particularly detrimental to the analysis of transient events: SN-like light-curves. We offer three main contributions: 1) Based on temporal modulation and attention mechanisms, we propose a Deep attention model (TimeModAttn) to classify multi-band light-curves of different SN types, avoiding photometric or hand-crafted feature computations, missing-value assumptions, and explicit imputation/interpolation methods. 2) We propose a model for the synthetic generation of SN multi-band light-curves based on the Supernova Parametric Model, allowing us to increase the number of samples and the diversity of cadence. Thus, the TimeModAttn model is first pre-trained using synthetic light-curves. Then, a fine-tuning process is performed. The TimeModAttn model outperformed other Deep Learning models, based on Recurrent Neural Networks, in two scenarios: late-classification and early-classification. Also, the TimeModAttn model outperformed a Balanced Random Forest (BRF) classifier (trained with real data), increasing the balanced-$F_1$score from $\approx.525$ to $\approx.596$. When training the BRF with synthetic data, this model achieved similar performance to the TimeModAttn model proposed while still maintaining extra advantages. 3) We conducted interpretability experiments. High attention scores were obtained for observations earlier than and close to the SN brightness peaks. This also correlated with an early highly variability of the learned temporal modulation.

Concentrated solar power (CSP) is one of the growing technologies that is leading the process of changing from fossil fuels to renewable energies. The sophistication and size of the systems require an increase in maintenance tasks to ensure reliability, availability, maintainability and safety. Currently, automatic fault detection in CSP plants using Parabolic Trough Collector systems evidences two main drawbacks: 1) the devices in use needs to be manually placed near the receiver tube, 2) the Machine Learning-based solutions are not tested in real plants. We address both gaps by combining the data extracted with the use of an Unmaned Aerial Vehicle, and the data provided by sensors placed within 7 real plants. The resulting dataset is the first one of this type and can help to standardize research activities for the problem of fault detection in this type of plants. Our work proposes supervised machine-learning algorithms for detecting broken envelopes of the absorber tubes in CSP plants. The proposed solution takes the class imbalance problem into account, boosting the accuracy of the algorithms for the minority class without harming the overall performance of the models. For a Deep Residual Network, we solve an imbalance and a balance problem at the same time, which increases by 5% the Recall of the minority class with no harm to the F1-score. Additionally, the Random Under Sampling technique boost the performance of traditional Machine Learning models, being the Histogram Gradient Boost Classifier the algorithm with the highest increase (3%) in the F1-Score. To the best of our knowledge, this paper is the first providing an automated solution to this problem using data from operating plants.

Objectives: Early detection of liver fibrosis can help cure the disease or prevent disease progression. We perform a comprehensive study of machine learning-based fibrosis detection in CT images using radiomic features to develop a non-invasive approach to fibrosis detection. Methods: Two sets of radiomic features were extracted from spherical ROIs in CT images of 182 patients who underwent simultaneous liver biopsy and CT examinations, one set corresponding to biopsy locations and another distant from biopsy locations. Combinations of contrast, normalization, machine learning model, feature selection method, bin width, and kernel radius were investigated, each of which were trained and evaluated 100 times with randomized development and test cohorts. The best settings were evaluated based on their mean test AUC and the best features were determined based on their frequency among the best settings. Results: Logistic regression models with NC images normalized using Gamma correction with $\gamma = 1.5$ performed best for fibrosis detection. Boruta was the best for radiomic feature selection method. Training a model using these optimal settings and features consisting of first order energy, first order kurtosis, and first order skewness, resulted in a model that achieved mean test AUCs of 0.7549 and 0.7166 on biopsy-based and non-biopsy ROIs respectively, outperforming a baseline and best models found during the initial study. Conclusions: Logistic regression models trained on radiomic features from NC images normalized using Gamma correction with $\gamma = 1.5$ that underwent Boruta feature selection are effective for liver fibrosis detection. Energy, kurtosis, and skewness are particularly effective features for fibrosis detection.

In this article, we propose using deep learning and transformer architectures combined with classical machine learning algorithms to detect and identify text anomalies in texts. Deep learning model provides a very crucial context information about the textual data which all textual context are converted to a numerical representation. We used multiple machine learning methods such as Sentence Transformers, Auto Encoders, Logistic Regression and Distance calculation methods to predict anomalies. The method are tested on the texts data and we used syntactic data from different source injected into the original text as anomalies or use them as target. Different methods and algorithm are explained in the field of outlier detection and the results of the best technique is presented. These results suggest that our algorithm could potentially reduce false positive rates compared with other anomaly detection methods that we are testing.

Quality control is a crucial activity performed by manufacturing enterprises to ensure that their products meet quality standards and avoid potential damage to the brand's reputation. The decreased cost of sensors and connectivity enabled increasing digitalization of manufacturing. In addition, artificial intelligence enables higher degrees of automation, reducing overall costs and time required for defect inspection. This research compares three active learning approaches, having single and multiple oracles, to visual inspection. Six new metrics are proposed to assess the quality of calibration without the need for ground truth. Furthermore, this research explores whether existing calibrators can improve their performance by leveraging an approximate ground truth to enlarge the calibration set. The experiments were performed on real-world data provided by Philips Consumer Lifestyle BV. Our results show that the explored active learning settings can reduce the data labeling effort by between three and four percent without detriment to the overall quality goals, considering a threshold of p=0.95. Furthermore, the results show that the proposed calibration metrics successfully capture relevant information otherwise available to metrics used up to date only through ground truth data. Therefore, the proposed metrics can be used to estimate the quality of models' probability calibration without committing to a labeling effort to obtain ground truth data.

Recognizing out-of-distribution (OOD) samples is critical for machine learning systems deployed in the open world. The vast majority of OOD detection methods are driven by a single modality (e.g., either vision or language), leaving the rich information in multi-modal representations untapped. Inspired by the recent success of vision-language pre-training, this paper enriches the landscape of OOD detection from a single-modal to a multi-modal regime. Particularly, we propose Maximum Concept Matching (MCM), a simple yet effective zero-shot OOD detection method based on aligning visual features with textual concepts. We contribute in-depth analysis and theoretical insights to understand the effectiveness of MCM. Extensive experiments demonstrate that MCM achieves superior performance on a wide variety of real-world tasks. MCM with vision-language features outperforms a common baseline with pure visual features on a hard OOD task with semantically similar classes by 13.1% (AUROC). Code is available at https://github.com/

Deep learning methods have boosted the adoption of NLP systems in real-life applications. However, they turn out to be vulnerable to distribution shifts over time which may cause severe dysfunctions in production systems, urging practitioners to develop tools to detect out-of-distribution (OOD) samples through the lens of the neural network. In this paper, we introduce TRUSTED, a new OOD detector for classifiers based on Transformer architectures that meets operational requirements: it is unsupervised and fast to compute. The efficiency of TRUSTED relies on the fruitful idea that all hidden layers carry relevant information to detect OOD examples. Based on this, for a given input, TRUSTED consists in (i) aggregating this information and (ii) computing a similarity score by exploiting the training distribution, leveraging the powerful concept of data depth. Our extensive numerical experiments involve 51k model configurations, including various checkpoints, seeds, and datasets, and demonstrate that TRUSTED achieves state-of-the-art performances. In particular, it improves previous AUROC over 3 points.

In anomaly detection, a prominent task is to induce a model to identify anomalies learned solely based on normal data. Generally, one is interested in finding an anomaly detector that correctly identifies anomalies, i.e., data points that do not belong to the normal class, without raising too many false alarms. Which anomaly detector is best suited depends on the dataset at hand and thus needs to be tailored. The quality of an anomaly detector may be assessed via confusion-based metrics such as the Matthews correlation coefficient (MCC). However, since during training only normal data is available in a semi-supervised setting, such metrics are not accessible. To facilitate automated machine learning for anomaly detectors, we propose to employ meta-learning to predict MCC scores based on metrics that can be computed with normal data only. First promising results can be obtained considering the hypervolume and the false positive rate as meta-features.