Search

Recent work in XAI for eye tracking data has evaluated the suitability of feature attribution methods to explain the output of deep neural sequence models for the task of oculomotric biometric identification. These methods provide saliency maps to highlight important input features of a specific eye gaze sequence. However, to date, its localization analysis has been lacking a quantitative approach across entire datasets. In this work, we employ established gaze event detection algorithms for fixations and saccades and quantitatively evaluate the impact of these events by determining their concept influence. Input features that belong to saccades are shown to be substantially more important than features that belong to fixations. By dissecting saccade events into sub-events, we are able to show that gaze samples that are close to the saccadic peak velocity are most influential. We further investigate the effect of event properties like saccadic amplitude or fixational dispersion on the resulting concept influence., Preprint for ETRA '23: 2023 Symposium on Eye Tracking Research and Applications

We introduce pymovements: a Python package for analyzing eye-tracking data that follows best practices in software development, including rigorous testing and adherence to coding standards. The package provides functionality for key processes along the entire preprocessing pipeline. This includes parsing of eye tracker data files, transforming positional data into velocity data, detecting gaze events like saccades and fixations, computing event properties like saccade amplitude and fixational dispersion and visualizing data and results with several types of plotting methods. Moreover, pymovements also provides an easily accessible interface for downloading and processing publicly available datasets. Additionally, we emphasize how rigorous testing in scientific software packages is critical to the reproducibility and transparency of research, enabling other researchers to verify and build upon previous findings., Preprint for ETRA '23: 2023 Symposium on Eye Tracking Research and Applications

We study involuntary micro-movements of both eyes, in addition to saccadic macro-movements, as biometric characteristic. We develop a deep convolutional neural network that processes binocular eye-tracking signals and verifies the viewer’s identity. In order to detect presentation attacks, we develop a model in which the movements are a response to a controlled stimulus. The model detects replay attacks by processing both the controlled but randomized stimulus and the ocular response to this stimulus. We acquire eye movement data from 150 participants, with 4 sessions per participant and conduct experiments on this new and legacy data sets with varying tracker precision and sampling rate. We observe that the model detects replay attacks reliably. For identification and identity verification, the model attains substantially lower error rates than prior work. We explore the relationships between training population size, training data volume, types of visual stimuli, number of training and enrollment sessions, interval between enrollment and probe sessions on one hand and the model performance on the other hand.

Computational drug sensitivity models have the potential to improve therapeutic outcomes by identifying targeted drug components that are likely to achieve the highest efficacy for a cancer cell line at hand at a therapeutic dose. State of the art drug sensitivity models use regression techniques to predict the inhibitory concentration of a drug for a tumor cell line. This regression objective is not directly aligned with either of these principal goals of drug sensitivity models: We argue that drug sensitivity modeling should be seen as a ranking problem with an optimization criterion that quantifies a drug’s inhibitory capacity for the cancer cell line at hand relative to its toxicity for healthy cells. We derive an extension to the well-established drug sensitivity regression model PaccMann that employs a ranking loss and focuses on the ratio of inhibitory concentration and therapeutic dosage range. We find that the ranking extension significantly enhances the model’s capability to identify the most effective anticancer drugs for unseen tumor cell profiles based in on in-vitro data.

Distributional properties of fixations and saccades are known to constitute biometric characteristics. Additionally, high-frequency micro-movements of the eyes have recently been found to constitute biometric characteristics that allow for faster and more robust biometric identification than just macro-movements. Micro-movements of the eyes occur on scales that are very close to the precision of currently available eye trackers. This study therefore characterizes the relationship between the temporal and spatial resolution of eye tracking recordings on one hand and the performance of a biometric identification method that processes micro-and macro-movements via a deep convolutional network. We find that that the deteriorating effects of decreasing both, the temporal and spatial resolution are not cumulative. We observe that on low-resolution data, the network reaches performance levels above chance and outperforms statistical approaches.

Background There is a huge body of scientific literature describing the relation between tumor types and anti-cancer drugs. The vast amount of scientific literature makes it impossible for researchers and physicians to extract all relevant information manually. Methods In order to cope with the large amount of literature we applied an automated text mining approach to assess the relations between 30 most frequent cancer types and 270 anti-cancer drugs. We applied two different approaches, a classical text mining based on named entity recognition and an AI-based approach employing word embeddings. The consistency of literature mining results was validated with 3 independent methods: first, using data from FDA approvals, second, using experimentally measured IC-50 cell line data and third, using clinical patient survival data. Results We demonstrated that the automated text mining was able to successfully assess the relation between cancer types and anti-cancer drugs. All validation methods showed a good correspondence between the results from literature mining and independent confirmatory approaches. The relation between most frequent cancer types and drugs employed for their treatment were visualized in a large heatmap. All results are accessible in an interactive web-based knowledge base using the following link: https://knowledgebase.microdiscovery.de/heatmap. Conclusions Our approach is able to assess the relations between compounds and cancer types in an automated manner. Both, cancer types and compounds could be grouped into different clusters. Researchers can use the interactive knowledge base to inspect the presented results and follow their own research questions, for example the identification of novel indication areas for known drugs.

We address the problems of identifying malware in network telemetry logs and providing indicators of compromise—comprehensible explanations of behavioral patterns that identify the threat. In our system, an array of specialized detectors abstracts network-flow data into comprehensible network events in a first step. We develop a neural network that processes this sequence of events and identifies specific threats, malware families and broad categories of malware. We then use the integrated-gradients method to highlight events that jointly constitute the characteristic behavioral pattern of the threat. We compare network architectures based on CNNs, LSTMs, and transformers, and explore the efficacy of unsupervised pre-training experimentally on large-scale telemetry data. We demonstrate how this system detects njRAT and other malware based on behavioral patterns.

We study involuntary micro-movements of both eyes, in addition to saccadic macro-movements, as biometric characteristic. We develop a deep convolutional neural network that processes binocular oculomotoric signals and identifies the viewer. In order to be able to detect presentation attacks, we develop a model in which the movements are a response to a controlled stimulus. The model detects replay attacks by processing both the controlled but randomized stimulus and the ocular response to this stimulus. We acquire eye movement data from 150 participants, with 4 sessions per participant. We observe that the model detects replay attacks reliably; compared to prior work, the model attains substantially lower error rates.

The temporal prediction of groundwater levels plays an important role in groundwater management, such as the estimation of anthropogenic impacts as well as consequences of climatic changes. Therefore, the modeling of groundwater levels using physics-based approaches is an integral part of hydrogeology. However, data-driven approaches have only recently been used, in particular for the prediction of groundwater levels using machine learning techniques (e.g., Random Forest and Neural Networks). For this purpose, one model per observation well or time series is always set up. In order to further develop this, an approach is presented which uses a single model for the prediction of groundwater levels at several observation wells (n > 200). The model is a three-dimensional Convolutional Neural Net (CNN).In addition to the time series of groundwater levels meteorological data on precipitation (P) and temperature (T) serves as additional input channels. The CNN "sees" not only the P- or T-value of the grid cell in which the observation well lies, but also surrounding values. This has the advantage that even influences of meteorological patterns in the spatial vicinity of the observation well on the groundwater level can be learned. The forecasts are calculated for periods up to six months. In addition to the comparison with the measured values, a comparison of the error averaged over all observation wells compared to a baseline model is used for the validation. To further improve predictability, the hyperparameters are optimized and other area data (e.g., geology, soil properties, land use) used as input. This model should form the basis for a regionalized forecast of groundwater levels.

Detection of malware-infected computers and detection of malicious web domains based on their encrypted HTTPS traffic are challenging problems, because only addresses, timestamps, and data volumes are observable. The detection problems are coupled, because infected clients tend to interact with malicious domains. Traffic data can be collected at a large scale, and antivirus tools can be used to identify infected clients in retrospect. Domains, by contrast, have to be labeled individually after forensic analysis. We explore transfer learning based on sluice networks; this allows the detection models to bootstrap each other. In a large-scale experimental study, we find that the model outperforms known reference models and detects previously unknown malware, previously unknown malware families, and previously unknown malicious domains., Comment: Mach Learn (2019)

In order to evade network-traffic analysis, an increasing proportion of malware uses the encrypted HTTPS protocol. We study the problem of detecting malware on client computers based on HTTPS traffic analysis. Here, malware has to be detected based on the host address, timestamps, and data volume information of aggregated packets that are sent and received by all the applications on the client. We develop a scalable protocol that allows us to collect network flows of known malicious and benign applications as training data and derive a malware-detection method based on a neural language model and a long short-term memory (LSTM) network. We study the method's ability to detect new malware in a large-scale empirical study.

We study the problem of detecting malware on client computers based on the analysis of HTTPS traffic. Here, malware has to be detected based on the host address, timestamps, and data volume information of the computer’s network traffic. We develop a scalable protocol that allows us to collect network flows of known malicious and benign applications as training data and derive a malware-detection method based on a neural embedding of domain names and a long short-term memory network that processes network flows. We study the method’s ability to detect new malware in a large-scale empirical study.