GANomaly is a semi-supervised anomaly detection method that uses adversarial training. The model employs a conditional generative adversarial network (GAN) to learn both the generation of high-dimensional image data and the inference of latent space representations. It uses an encoder-decoder-encoder architecture to map input images to a lower-dimensional vector, which is then used to reconstruct the image. The additional encoder network maps the generated image to its latent representation, helping the model learn the data distribution of normal samples. During training, minimizing the distance between generated images and latent vectors helps the model learn the normal data distribution. At inference, a larger distance from this distribution indicates an anomaly. The model is tested on various benchmark datasets, including MNIST, CIFAR, and X-ray security screening data. It outperforms previous state-of-the-art approaches in terms of both statistical and computational performance. The model is efficient and can be applied to a wide range of anomaly detection tasks. It uses adversarial training to learn the data distribution of normal samples and detect anomalies by measuring the distance between the generated image and its latent representation. The model is trained on normal samples and tested on both normal and abnormal samples. The anomaly score is calculated based on the distance between the generated image and its latent representation. The model's performance is evaluated using metrics such as the area under the curve (AUC) of the receiver operating characteristic (ROC) and true positive rate (TPR) as a function of false positive rate (FPR). The results show that GANomaly achieves superior performance compared to other methods in detecting anomalies in various datasets.GANomaly is a semi-supervised anomaly detection method that uses adversarial training. The model employs a conditional generative adversarial network (GAN) to learn both the generation of high-dimensional image data and the inference of latent space representations. It uses an encoder-decoder-encoder architecture to map input images to a lower-dimensional vector, which is then used to reconstruct the image. The additional encoder network maps the generated image to its latent representation, helping the model learn the data distribution of normal samples. During training, minimizing the distance between generated images and latent vectors helps the model learn the normal data distribution. At inference, a larger distance from this distribution indicates an anomaly. The model is tested on various benchmark datasets, including MNIST, CIFAR, and X-ray security screening data. It outperforms previous state-of-the-art approaches in terms of both statistical and computational performance. The model is efficient and can be applied to a wide range of anomaly detection tasks. It uses adversarial training to learn the data distribution of normal samples and detect anomalies by measuring the distance between the generated image and its latent representation. The model is trained on normal samples and tested on both normal and abnormal samples. The anomaly score is calculated based on the distance between the generated image and its latent representation. The model's performance is evaluated using metrics such as the area under the curve (AUC) of the receiver operating characteristic (ROC) and true positive rate (TPR) as a function of false positive rate (FPR). The results show that GANomaly achieves superior performance compared to other methods in detecting anomalies in various datasets.