Jörn Hees

Brian Moser, Federico Raue, Stanislav Frolov et al.

With the advent of Deep Learning (DL), Super-Resolution (SR) has also become a thriving research area. However, despite promising results, the field still faces challenges that require further research e.g., allowing flexible upsampling, more effective loss functions, and better evaluation metrics. We review the domain of SR in light of recent advances, and examine state-of-the-art models such as diffusion (DDPM) and transformer-based SR models. We present a critical discussion on contemporary strategies used in SR, and identify promising yet unexplored research directions. We complement previous surveys by incorporating the latest developments in the field such as uncertainty-driven losses, wavelet networks, neural architecture search, novel normalization methods, and the latests evaluation techniques. We also include several visualizations for the models and methods throughout each chapter in order to facilitate a global understanding of the trends in the field. This review is ultimately aimed at helping researchers to push the boundaries of DL applied to SR.

Brian Moser, Federico Raue, Jörn Hees et al.

Neural Architecture Search (NAS) defines the design of Neural Networks as a search problem. Unfortunately, NAS is computationally intensive because of various possibilities depending on the number of elements in the design and the possible connections between them. In this work, we extensively analyze the role of the dataset size based on several sampling approaches for reducing the dataset size (unsupervised and supervised cases) as an agnostic approach to reduce search time. We compared these techniques with four common NAS approaches in NAS-Bench-201 in roughly 1,400 experiments on CIFAR-100. One of our surprising findings is that in most cases we can reduce the amount of training data to 25\%, consequently reducing search time to 25\%, while at the same time maintaining the same accuracy as if training on the full dataset. Additionally, some designs derived from subsets out-perform designs derived from the full dataset by up to 22 p.p. accuracy.

Brian Moser, Federico Raue, Jörn Hees et al.

We present new Recurrent Neural Network (RNN) cells for image classification using a Neural Architecture Search (NAS) approach called DARTS. We are interested in the ReNet architecture, which is a RNN based approach presented as an alternative for convolutional and pooling steps. ReNet can be defined using any standard RNN cells, such as LSTM and GRU. One limitation is that standard RNN cells were designed for one dimensional sequential data and not for two dimensions like it is the case for image classification. We overcome this limitation by using DARTS to find new cell designs. We compare our results with ReNet that uses GRU and LSTM cells. Our found cells outperform the standard RNN cells on CIFAR-10 and SVHN. The improvements on SVHN indicate generalizability, as we derived the RNN cell designs from CIFAR-10 without performing a new cell search for SVHN.

Stanislav Frolov, Prateek Bansal, Jörn Hees et al.

Despite astonishing progress, generating realistic images of complex scenes remains a challenging problem. Recently, layout-to-image synthesis approaches have attracted much interest by conditioning the generator on a list of bounding boxes and corresponding class labels. However, previous approaches are very restrictive because the set of labels is fixed a priori. Meanwhile, text-to-image synthesis methods have substantially improved and provide a flexible way for conditional image generation. In this work, we introduce dense text-to-image (DT2I) synthesis as a new task to pave the way toward more intuitive image generation. Furthermore, we propose DTC-GAN, a novel method to generate images from semantically rich region descriptions, and a multi-modal region feature matching loss to encourage semantic image-text matching. Our results demonstrate the capability of our approach to generate plausible images of complex scenes using region captions.

Timur Sattarov, Dayananda Herurkar, Jörn Hees

Recent advances in Explainable AI (XAI) increased the demand for deployment of safe and interpretable AI models in various industry sectors. Despite the latest success of deep neural networks in a variety of domains, understanding the decision-making process of such complex models still remains a challenging task for domain experts. Especially in the financial domain, merely pointing to an anomaly composed of often hundreds of mixed type columns, has limited value for experts. Hence, in this paper, we propose a framework for explaining anomalies using denoising autoencoders designed for mixed type tabular data. We specifically focus our technique on anomalies that are erroneous observations. This is achieved by localizing individual sample columns (cells) with potential errors and assigning corresponding confidence scores. In addition, the model provides the expected cell value estimates to fix the errors. We evaluate our approach based on three standard public tabular datasets (Credit Default, Adult, IEEE Fraud) and one proprietary dataset (Holdings). We find that denoising autoencoders applied to this task already outperform other approaches in the cell error detection rates as well as in the expected value rates. Additionally, we analyze how a specialized loss designed for cell error detection can further improve these metrics. Our framework is designed for a domain expert to understand abnormal characteristics of an anomaly, as well as to improve in-house data quality management processes.

Mohammad Majd Saad Al Deen, Maren Pielka, Jörn Hees et al.

This paper addresses the classification of Arabic text data in the field of Natural Language Processing (NLP), with a particular focus on Natural Language Inference (NLI) and Contradiction Detection (CD). Arabic is considered a resource-poor language, meaning that there are few data sets available, which leads to limited availability of NLP methods. To overcome this limitation, we create a dedicated data set from publicly available resources. Subsequently, transformer-based machine learning models are being trained and evaluated. We find that a language-specific model (AraBERT) performs competitively with state-of-the-art multilingual approaches, when we apply linguistically informed pre-training methods such as Named Entity Recognition (NER). To our knowledge, this is the first large-scale evaluation for this task in Arabic, as well as the first application of multi-task pre-training in this context.

Stanislav Frolov, Avneesh Sharma, Jörn Hees et al.

Conditional image synthesis from layout has recently attracted much interest. Previous approaches condition the generator on object locations as well as class labels but lack fine-grained control over the diverse appearance aspects of individual objects. Gaining control over the image generation process is fundamental to build practical applications with a user-friendly interface. In this paper, we propose a method for attribute controlled image synthesis from layout which allows to specify the appearance of individual objects without affecting the rest of the image. We extend a state-of-the-art approach for layout-to-image generation to additionally condition individual objects on attributes. We create and experiment on a synthetic, as well as the challenging Visual Genome dataset. Our qualitative and quantitative results show that our method can successfully control the fine-grained details of individual objects when modelling complex scenes with multiple objects. Source code, dataset and pre-trained models are publicly available (https://github.com/stanifrolov/AttrLostGAN).

Dayananda Herurkar, Jörn Hees, Vesselin Tzvetkov et al.

The remarkable success of Deep Learning approaches is often based and demonstrated on large public datasets. However, when applying such approaches to internal, private datasets, one frequently faces challenges arising from structural differences in the datasets, domain shift, and the lack of labels. In this work, we introduce Tabular Data Adapters (TDA), a novel method for generating soft labels for unlabeled tabular data in outlier detection tasks. By identifying statistically similar public datasets and transforming private data (based on a shared autoencoder) into a format compatible with state-of-the-art public models, our approach enables the generation of weak labels. It thereby can help to mitigate the cold start problem of labeling by basing on existing outlier detection models for public datasets. In experiments on 50 tabular datasets across different domains, we demonstrate that our method is able to provide more accurate annotations than baseline approaches while reducing computational time. Our approach offers a scalable, efficient, and cost-effective solution, to bridge the gap between public research models and real-world industrial applications.

Fatemeh Azimi, Jean-Francois Jacques Nicolas Nies, Sebastian Palacio et al.

Curriculum learning is a bio-inspired training technique that is widely adopted to machine learning for improved optimization and better training of neural networks regarding the convergence rate or obtained accuracy. The main concept in curriculum learning is to start the training with simpler tasks and gradually increase the level of difficulty. Therefore, a natural question is how to determine or generate these simpler tasks. In this work, we take inspiration from Spatial Transformer Networks (STNs) in order to form an easy-to-hard curriculum. As STNs have been proven to be capable of removing the clutter from the input images and obtaining higher accuracy in image classification tasks, we hypothesize that images processed by STNs can be seen as easier tasks and utilized in the interest of curriculum learning. To this end, we study multiple strategies developed for shaping the training curriculum, using the data generated by STNs. We perform various experiments on cluttered MNIST and Fashion-MNIST datasets, where on the former, we obtain an improvement of $3.8$pp in classification accuracy compared to the baseline.

Andrey Guzhov, Federico Raue, Jörn Hees et al.

In the past, the rapidly evolving field of sound classification greatly benefited from the application of methods from other domains. Today, we observe the trend to fuse domain-specific tasks and approaches together, which provides the community with new outstanding models. In this work, we present an extension of the CLIP model that handles audio in addition to text and images. Our proposed model incorporates the ESResNeXt audio-model into the CLIP framework using the AudioSet dataset. Such a combination enables the proposed model to perform bimodal and unimodal classification and querying, while keeping CLIP's ability to generalize to unseen datasets in a zero-shot inference fashion. AudioCLIP achieves new state-of-the-art results in the Environmental Sound Classification (ESC) task, out-performing other approaches by reaching accuracies of 90.07% on the UrbanSound8K and 97.15% on the ESC-50 datasets. Further it sets new baselines in the zero-shot ESC-task on the same datasets (68.78% and 69.40%, respectively). Finally, we also assess the cross-modal querying performance of the proposed model as well as the influence of full and partial training on the results. For the sake of reproducibility, our code is published.

Ricard Durall, Stanislav Frolov, Jörn Hees et al.

Transformer models have recently attracted much interest from computer vision researchers and have since been successfully employed for several problems traditionally addressed with convolutional neural networks. At the same time, image synthesis using generative adversarial networks (GANs) has drastically improved over the last few years. The recently proposed TransGAN is the first GAN using only transformer-based architectures and achieves competitive results when compared to convolutional GANs. However, since transformers are data-hungry architectures, TransGAN requires data augmentation, an auxiliary super-resolution task during training, and a masking prior to guide the self-attention mechanism. In this paper, we study the combination of a transformer-based generator and convolutional discriminator and successfully remove the need of the aforementioned required design choices. We evaluate our approach by conducting a benchmark of well-known CNN discriminators, ablate the size of the transformer-based generator, and show that combining both architectural elements into a hybrid model leads to better results. Furthermore, we investigate the frequency spectrum properties of generated images and observe that our model retains the benefits of an attention based generator.

Sebastian Palacio, Adriano Lucieri, Mohsin Munir et al.

The field of explainable AI (XAI) has quickly become a thriving and prolific community. However, a silent, recurrent and acknowledged issue in this area is the lack of consensus regarding its terminology. In particular, each new contribution seems to rely on its own (and often intuitive) version of terms like "explanation" and "interpretation". Such disarray encumbers the consolidation of advances in the field towards the fulfillment of scientific and regulatory demands e.g., when comparing methods or establishing their compliance with respect to biases and fairness constraints. We propose a theoretical framework that not only provides concrete definitions for these terms, but it also outlines all steps necessary to produce explanations and interpretations. The framework also allows for existing contributions to be re-contextualized such that their scope can be measured, thus making them comparable to other methods. We show that this framework is compliant with desiderata on explanations, on interpretability and on evaluation metrics. We present a use-case showing how the framework can be used to compare LIME, SHAP and MDNet, establishing their advantages and shortcomings. Finally, we discuss relevant trends in XAI as well as recommendations for future work, all from the standpoint of our framework.

Andrey Guzhov, Federico Raue, Jörn Hees et al.

Environmental Sound Classification (ESC) is a rapidly evolving field that recently demonstrated the advantages of application of visual domain techniques to the audio-related tasks. Previous studies indicate that the domain-specific modification of cross-domain approaches show a promise in pushing the whole area of ESC forward. In this paper, we present a new time-frequency transformation layer that is based on complex frequency B-spline (fbsp) wavelets. Being used with a high-performance audio classification model, the proposed fbsp-layer provides an accuracy improvement over the previously used Short-Time Fourier Transform (STFT) on standard datasets. We also investigate the influence of different pre-training strategies, including the joint use of two large-scale datasets for weight initialization: ImageNet and AudioSet. Our proposed model out-performs other approaches by achieving accuracies of 95.20 % on the ESC-50 and 89.14 % on the UrbanSound8K datasets. Additionally, we assess the increase of model robustness against additive white Gaussian noise and reduction of an effective sample rate introduced by the proposed layer and demonstrate that the fbsp-layer improves the model's ability to withstand signal perturbations, in comparison to STFT-based training. For the sake of reproducibility, our code is made available.

Jörn Hees, Dayananda Herurkar, Mario Meier

Detecting outliers or anomalies is a common data analysis task. As a sub-field of unsupervised machine learning, a large variety of approaches exist, but the vast majority treats the input features as independent and often fails to recognize even simple (linear) relationships in the input feature space. Hence, we introduce RECol, a generic data pre-processing approach to generate additional columns in a leave-one-out-fashion: For each column, we try to predict its values based on the other columns, generating reconstruction error columns. We run experiments across a large variety of common baseline approaches and benchmark datasets with and without our RECol pre-processing method and show that the generated reconstruction error feature space generally seems to support common outlier detection methods and often considerably improves their ROC-AUC and PR-AUC values.

Stanislav Frolov, Tobias Hinz, Federico Raue et al.

With the advent of generative adversarial networks, synthesizing images from textual descriptions has recently become an active research area. It is a flexible and intuitive way for conditional image generation with significant progress in the last years regarding visual realism, diversity, and semantic alignment. However, the field still faces several challenges that require further research efforts such as enabling the generation of high-resolution images with multiple objects, and developing suitable and reliable evaluation metrics that correlate with human judgement. In this review, we contextualize the state of the art of adversarial text-to-image synthesis models, their development since their inception five years ago, and propose a taxonomy based on the level of supervision. We critically examine current strategies to evaluate text-to-image synthesis models, highlight shortcomings, and identify new areas of research, ranging from the development of better datasets and evaluation metrics to possible improvements in architectural design and model training. This review complements previous surveys on generative adversarial networks with a focus on text-to-image synthesis which we believe will help researchers to further advance the field.

Sebastian Palacio, Philipp Engler, Jörn Hees et al.

Classification problems solved with deep neural networks (DNNs) typically rely on a closed world paradigm, and optimize over a single objective (e.g., minimization of the cross-entropy loss). This setup dismisses all kinds of supporting signals that can be used to reinforce the existence or absence of a particular pattern. The increasing need for models that are interpretable by design makes the inclusion of said contextual signals a crucial necessity. To this end, we introduce the notion of Self-Supervised Autogenous Learning (SSAL) models. A SSAL objective is realized through one or more additional targets that are derived from the original supervised classification task, following architectural principles found in multi-task learning. SSAL branches impose low-level priors into the optimization process (e.g., grouping). The ability of using SSAL branches during inference, allow models to converge faster, focusing on a richer set of class-relevant features. We show that SSAL models consistently outperform the state-of-the-art while also providing structured predictions that are more interpretable.

Stanislav Frolov, Shailza Jolly, Jörn Hees et al.

Generating images from textual descriptions has recently attracted a lot of interest. While current models can generate photo-realistic images of individual objects such as birds and human faces, synthesising images with multiple objects is still very difficult. In this paper, we propose an effective way to combine Text-to-Image (T2I) synthesis with Visual Question Answering (VQA) to improve the image quality and image-text alignment of generated images by leveraging the VQA 2.0 dataset. We create additional training samples by concatenating question and answer (QA) pairs and employ a standard VQA model to provide the T2I model with an auxiliary learning signal. We encourage images generated from QA pairs to look realistic and additionally minimize an external VQA loss. Our method lowers the FID from 27.84 to 25.38 and increases the R-prec. from 83.82% to 84.79% when compared to the baseline, which indicates that T2I synthesis can successfully be improved using a standard VQA model.

Andrey Guzhov, Federico Raue, Jörn Hees et al.

Environmental Sound Classification (ESC) is an active research area in the audio domain and has seen a lot of progress in the past years. However, many of the existing approaches achieve high accuracy by relying on domain-specific features and architectures, making it harder to benefit from advances in other fields (e.g., the image domain). Additionally, some of the past successes have been attributed to a discrepancy of how results are evaluated (i.e., on unofficial splits of the UrbanSound8K (US8K) dataset), distorting the overall progression of the field. The contribution of this paper is twofold. First, we present a model that is inherently compatible with mono and stereo sound inputs. Our model is based on simple log-power Short-Time Fourier Transform (STFT) spectrograms and combines them with several well-known approaches from the image domain (i.e., ResNet, Siamese-like networks and attention). We investigate the influence of cross-domain pre-training, architectural changes, and evaluate our model on standard datasets. We find that our model out-performs all previously known approaches in a fair comparison by achieving accuracies of 97.0 % (ESC-10), 91.5 % (ESC-50) and 84.2 % / 85.4 % (US8K mono / stereo). Second, we provide a comprehensive overview of the actual state of the field, by differentiating several previously reported results on the US8K dataset between official or unofficial splits. For better reproducibility, our code (including any re-implementations) is made available.

Philipp Blandfort, Tushar Karayil, Federico Raue et al.

Growing amounts of online user data motivate the need for automated processing techniques. In case of user ratings, one interesting option is to use neural networks for learning to predict ratings given an item and a user. While training for prediction, such an approach at the same time learns to map each user to a vector, a so-called user embedding. Such embeddings can for example be valuable for estimating user similarity. However, there are various ways how item and user information can be combined in neural networks, and it is unclear how the way of combining affects the resulting embeddings. In this paper, we run an experiment on movie ratings data, where we analyze the effect on embedding quality caused by several fusion strategies in neural networks. For evaluating embedding quality, we propose a novel measure, Pair-Distance Correlation, which quantifies the condition that similar users should have similar embedding vectors. We find that the fusion strategy affects results in terms of both prediction performance and embedding quality. Surprisingly, we find that prediction performance not necessarily reflects embedding quality. This suggests that if embeddings are of interest, the common tendency to select models based on their prediction ability should be reconsidered.

Philipp Blandfort, Jörn Hees, Desmond U. Patton

It is said that beauty is in the eye of the beholder. But how exactly can we characterize such discrepancies in interpretation? For example, are there any specific features of an image that makes person A regard an image as beautiful while person B finds the same image displeasing? Such questions ultimately aim at explaining our individual ways of interpretation, an intention that has been of fundamental importance to the social sciences from the beginning. More recently, advances in computer science brought up two related questions: First, can computational tools be adopted for analyzing ways of interpretation? Second, what if the "beholder" is a computer model, i.e., how can we explain a computer model's point of view? Numerous efforts have been made regarding both of these points, while many existing approaches focus on particular aspects and are still rather separate. With this paper, in order to connect these approaches we introduce a theoretical framework for analyzing interpretation, which is applicable to interpretation of both human beings and computer models. We give an overview of relevant computational approaches from various fields, and discuss the most common and promising application areas. The focus of this paper lies on interpretation of text and image data, while many of the presented approaches are applicable to other types of data as well.

Tushar Karayil, Philipp Blandfort, Jörn Hees et al.

Subjective visual interpretation is a challenging yet important topic in computer vision. Many approaches reduce this problem to the prediction of adjective- or attribute-labels from images. However, most of these do not take attribute semantics into account, or only process the image in a holistic manner. Furthermore, there is a lack of relevant datasets with fine-grained subjective labels. In this paper, we propose the Focus-Aspect-Polarity model to structure the process of capturing subjectivity in image processing, and introduce a novel dataset following this way of modeling. We run experiments on this dataset to compare several deep learning methods and find that incorporating context information based on tensor multiplication in several cases outperforms the default way of information fusion (concatenation).

Markus Schröder, Christian Jilek, Jörn Hees et al.

In the field of machine learning, data understanding is the practice of getting initial insights in unknown datasets. Such knowledge-intensive tasks require a lot of documentation, which is necessary for data scientists to grasp the meaning of the data. Usually, documentation is separate from the data in various external documents, diagrams, spreadsheets and tools which causes considerable look up overhead. Moreover, other supporting applications are not able to consume and utilize such unstructured data. That is why we propose a methodology that uses a single semantic model that interlinks data with its documentation. Hence, data scientists are able to directly look up the connected information about the data by simply following links. Equally, they can browse the documentation which always refers to the data. Furthermore, the model can be used by other approaches providing additional support, like searching, comparing, integrating or visualizing data. To showcase our approach we also demonstrate an early prototype.

Markus Schröder, Jörn Hees, Ansgar Bernardi et al.

Within the Semantic Web community, SPARQL is one of the predominant languages to query and update RDF knowledge. However, the complexity of SPARQL, the underlying graph structure and various encodings are common sources of confusion for Semantic Web novices. In this paper we present a general purpose approach to convert any given SPARQL endpoint into a simple to use REST API. To lower the initial hurdle, we represent the underlying graph as an interlinked view of nested JSON objects that can be traversed by the API path.

Markus Schröder, Christian Jilek, Jörn Hees et al.

Spreadsheets are widely used by knowledge workers, especially in the industrial sector. Their methodology enables a well understood, easy and fast possibility to enter data. As filling out a spreadsheet is more accessible to common knowledge workers than defining RDF statements, in this paper, we propose an easy-to-use, zero-configuration, web-based spreadsheet editor that simultaneously transfers spreadsheet entries into RDF statements. It enables various kinds of users to easily create semantic data whether they are RDF experts or novices. The typical scenario we address focuses on creating instance data starting with an empty knowledge base that is filled incrementally. In a user study, participants were able to create more statements in shorter time, having similar or even significantly outperforming quality, compared to other approaches.

Sebastian Palacio, Joachim Folz, Jörn Hees et al.

We propose a novel way to measure and understand convolutional neural networks by quantifying the amount of input signal they let in. To do this, an autoencoder (AE) was fine-tuned on gradients from a pre-trained classifier with fixed parameters. We compared the reconstructed samples from AEs that were fine-tuned on a set of image classifiers (AlexNet, VGG16, ResNet-50, and Inception~v3) and found substantial differences. The AE learns which aspects of the input space to preserve and which ones to ignore, based on the information encoded in the backpropagated gradients. Measuring the changes in accuracy when the signal of one classifier is used by a second one, a relation of total order emerges. This order depends directly on each classifier's input signal but it does not correlate with classification accuracy or network size. Further evidence of this phenomenon is provided by measuring the normalized mutual information between original images and auto-encoded reconstructions from different fine-tuned AEs. These findings break new ground in the area of neural network understanding, opening a new way to reason, debug, and interpret their results. We present four concrete examples in the literature where observations can now be explained in terms of the input signal that a model uses.

Jörn Hees, Rouven Bauer, Joachim Folz et al.

Efficient usage of the knowledge provided by the Linked Data community is often hindered by the need for domain experts to formulate the right SPARQL queries to answer questions. For new questions they have to decide which datasets are suitable and in which terminology and modelling style to phrase the SPARQL query. In this work we present an evolutionary algorithm to help with this challenging task. Given a training list of source-target node-pair examples our algorithm can learn patterns (SPARQL queries) from a SPARQL endpoint. The learned patterns can be visualised to form the basis for further investigation, or they can be used to predict target nodes for new source nodes. Amongst others, we apply our algorithm to a dataset of several hundred human associations (such as "circle - square") to find patterns for them in DBpedia. We show the scalability of the algorithm by running it against a SPARQL endpoint loaded with > 7.9 billion triples. Further, we use the resulting SPARQL queries to mimic human associations with a Mean Average Precision (MAP) of 39.9 % and a Recall@10 of 63.9 %.

Tim Althoff, Damian Borth, Jörn Hees et al.

Among the vast information available on the web, social media streams capture what people currently pay attention to and how they feel about certain topics. Awareness of such trending topics plays a crucial role in multimedia systems such as trend aware recommendation and automatic vocabulary selection for video concept detection systems. Correctly utilizing trending topics requires a better understanding of their various characteristics in different social media streams. To this end, we present the first comprehensive study across three major online and social media streams, Twitter, Google, and Wikipedia, covering thousands of trending topics during an observation period of an entire year. Our results indicate that depending on one's requirements one does not necessarily have to turn to Twitter for information about current events and that some media streams strongly emphasize content of specific categories. As our second key contribution, we further present a novel approach for the challenging task of forecasting the life cycle of trending topics in the very moment they emerge. Our fully automated approach is based on a nearest neighbor forecasting technique exploiting our assumption that semantically similar topics exhibit similar behavior. We demonstrate on a large-scale dataset of Wikipedia page view statistics that forecasts by the proposed approach are about 9-48k views closer to the actual viewing statistics compared to baseline methods and achieve a mean average percentage error of 45-19% for time periods of up to 14 days.