检索结果-内蒙古大学图书馆

Phantom Embeddings: Using Embedding Space for Model Regularization in Deep Neural Networks

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Arif, Mofassir ul Islam Jameel, Mohsan Grabocka, Josif Schmidt-Thieme, Lars Information Systems and Machine Learning Lab University of Hildesheim Germany Department for Computer Science Albert-Ludwigs-University Freiburg Germany

The strength of machine learning models stems from their ability to learn complex function approximations from data;however, this strength also makes training deep neural networks challenging. Notably, the complex models tend to memorize the training data, which results in poor regularization performance on test data. The regularization techniques such as L1, L2, dropout, etc. are proposed to reduce the overfitting effect;however, they bring in additional hyperparameters tuning complexity. These methods also fall short when the inter-class similarity is high due to the underlying data distribution, leading to a less accurate model. In this paper, we present a novel approach to regularize the models by leveraging the information-rich latent embeddings and their high intra-class correlation. We create phantom embeddings from a subset of homogenous samples and use these phantom embeddings to decrease the inter-class similarity of instances in their latent embedding space. The resulting models generalize better as a combination of their embedding, regularizes them without requiring an expensive hyperparameter search. We evaluate our method on two popular and challenging image classification datasets (CIFAR and FashionMNIST) and show how our approach outperforms the standard baselines while displaying better training behavior. © 2023, CC BY.

关键词： Deep neural networks

Offline Handwritten Amharic Character Recognition Using Few-Shot learning 1st

学校读者我要写书评

暂无评论

Offline Handwritten Amharic Character Recognition Using Few-...

1st Pan-African Conference on Artificial Intelligence, PanAfriCon AI 2022

作者： Samuel, Mesay Schmidt-Thieme, Lars Sharma, D.P. Sinamo, Abiot Bruck, Abey Faculty of Computing and Software Engineering Arba Minch University Arba Minch Ethiopia Information Systems and Machine Learning Lab Hildesheim31141 Germany AMUIT MOEFDRE under UNDP Addis Ababa Ethiopia Ministry of Innovation and Technology Federal Democratic Republic of Ethiopia Addis Ababa Ethiopia

ISBN: (纸本)9783031313264

Few-shot learning is a demanding and challenging problem in machine learning aimed at learning from only fewer labeled training examples. It has become an active area of research due to deep learning requiring huge amounts of labeled dataset, which is not easy practically. learning from fewer examples is also an important attempt towards learning like humans. Few-shot learning has proven a very good promise in different areas of machine learning applications, particularly in image classification. As it is a recent technique, most researchers focus on understanding and solving the issues related to its concept by focusing only on common image datasets like Mini-ImageNet and Omniglot. Few-shot learning also opens an opportunity to address low resource languages like Amharic. Hence in the current study, it is used to address offline handwritten character recognition for Amharic language. Particularly, prototypical networks, the popular and simpler type of few-shot learning, is implemented as a baseline. Using the opportunities explored in the nature of Amharic alphabet having columnwise and rowwise characters’ similarities, a novel way of augmenting the training episodes is proposed. According to the study’s experimental result, it was observed that the proposed method outperformed the baseline method in 1-shot setting. This study has implemented few-shot learning for Amharic characters for the first time. More importantly, the findings of the study open new ways of examining the influence of training episodes in few-shot learning, which is one of the critical issues requiring further exploration. The datasets used for this study are collected from native Amharic language writers using an Android App developed as a part of this study. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Character recognition

CARCA: Context and Attribute-Aware Next-Item Recommendation via Cross-Attention

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Rashed, Ahmed Elsayed, Shereen Schmidt-Thieme, Lars Information Systems and Machine Learning Lab University of Hildesheim Germany University of Hildesheim Germany

In sparse recommender settings, users' context and item attributes play a crucial role in deciding which items to recommend next. Despite that, recent works in sequential and time-aware recommendations usually either ignore both aspects or only consider one of them, limiting their predictive performance. In this paper, we address these limitations by proposing a context and attributeaware recommender model (CARCA) that can capture the dynamic nature of the user profiles in terms of contextual features and item attributes via dedicated multi-head self-attention blocks that extract profile-level features and predicting item scores. Also, unlike many of the current state-of-the-art sequential item recommendation approaches that use a simple dot-product between the most recent item's latent features and the target items embeddings for scoring, CARCA uses cross-attention between all profile items and the target items to predict their final scores. This cross-attention allows CARCA to harness the correlation between old and recent items in the user profile and their influence on deciding which item to recommend next. Experiments on four real-world recommender system datasets show that the proposed model significantly outperforms all state-of-the-art models in the task of item recommendation and achieving improvements of up to 53% in Normalized Discounted Cumulative Gain (NDCG) and Hit-Ratio. Results also show that CARCA outperformed several state-of-the-art dedicated imagebased recommender systems by merely utilizing image attributes extracted from a pre-trained ResNet50 in a black-box fashion. Copyright © 2022, The Authors. All rights reserved.

关键词： User profile

Latent Action Priors for Locomotion with Deep Reinforcement learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Hausdörfer, Oliver von Rohr, Alexander Lefort, Éric Schoellig, Angela P. The Technical University of Munich Germany TUM School of Computation Information and Technology Department of Computer Engineering Learning Systems and Robotics Lab Germany Munich Institute of Robotics and Machine Intelligence Germany

Deep Reinforcement learning (DRL) enables robots to learn complex behaviors through interaction with the environment. However, due to the unrestricted nature of the learning algorithms, the resulting solutions are often brittle and appear unnatural. This is especially true for learning direct joint-level torque control, as inductive biases are difficult to integrate into the learning process. We propose an inductive bias for learning locomotion that is especially useful for torque control: latent actions learned from a small dataset of expert demonstrations. This prior allows the policy to directly leverage knowledge contained in the expert’s actions and facilitates more efficient exploration. We observe that the agent is not restricted to the reward levels of the demonstration, and performance in transfer tasks is improved significantly. Latent action priors combined with style rewards for imitation leads to a closer replication of the expert’s behavior. Videos and code are available at https://***/view/ latent-action-priors. Copyright © 2024, The Authors. All rights reserved.

关键词： Deep reinforcement learning

ProbSAINT: Probabilistic Tabular Regression for Used Car Pricing

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Madhusudhanan, Kiran Behrens, Gunnar Stubbemann, Maximilian Schmidt-Thieme, Lars Information Systems and Machine Learning Lab VWFS Data Analytics Research Center University of Hildesheim Hildesheim Germany Data Analytics & AI Engineering Volkswagen Financial Services AG Braunschweig Germany

Used car pricing is a critical aspect of the automotive industry, influenced by many economic factors and market dynamics. With the recent surge in online marketplaces and increased demand for used cars, accurate pricing would benefit both buyers and sellers by ensuring fair transactions. However, the transition towards automated pricing algorithms using machine learning necessitates the comprehension of model uncertainties, specifically the ability to flag predictions that the model is unsure about. Although recent literature proposes the use of boosting algorithms or nearest neighbor-based approaches for swift and precise price predictions, encapsulating model uncertainties with such algorithms presents a complex challenge. We introduce ProbSAINT, a model that offers a principled approach for uncertainty quantification of its price predictions, along with accurate point predictions that are comparable to state-of-the-art boosting techniques. Furthermore, acknowledging that the business prefers pricing used cars based on the number of days the vehicle was listed for sale, we show how ProbSAINT can be used as a dynamic forecasting model for predicting price probabilities for different expected offer duration. Our experiments further indicate that ProbSAINT is especially accurate on instances where it is highly certain. This proves the applicability of its probabilistic predictions in real-world scenarios where trustworthiness is crucial. © 2024, CC BY.

关键词： Costs

Control-Barrier-Aided Teleoperation with Visual-Inertial SLAM for Safe MAV Navigation in Complex Environments

学校读者我要写书评

暂无评论

Control-Barrier-Aided Teleoperation with Visual-Inertial SLA...

IEEE International Conference on Robotics and Automation (ICRA)

作者： Siqi Zhou Sotiris Papatheodorou Stefan Leutenegger Angela P. Schoellig Learning Systems and Robotics Lab School of Computation Information and Technology Technical University of Munich Munich Institute of Robotics and Machine Intelligence (MIRMI) Smart Robotics Lab School of Computation Information and Technology Technical University of Munich Department of Computing Smart Robotics Lab Imperial College London

ISBN: (数字)9798350384574

ISBN: (纸本)9798350384581

In this paper, we consider a Micro Aerial Vehicle (MAV) system teleoperated by a non-expert and introduce a perceptive safety filter that leverages Control Barrier Functions (CBFs) in conjunction with Visual-Inertial Simultaneous Localization and Mapping (VI-SLAM) and dense 3D occupancy mapping to guarantee safe navigation in complex and unstructured environments. Our system relies solely on onboard IMU measurements, stereo infrared images, and depth images and autonomously corrects teleoperated inputs when they are deemed unsafe. We define a point in 3D space as unsafe if it satisfies either of two conditions: (i) it is occupied by an obstacle, or (ii) it remains unmapped. At each time step, an occupancy map of the environment is updated by the VI-SLAM by fusing the onboard measurements, and a CBF is constructed to parameterize the (un)safe region in the 3D space. Given the CBF and state feedback from the VI-SLAM module, a safety filter computes a certified reference that best matches the teleoperation input while satisfying the safety constraint encoded by the CBF. In contrast to existing perception-based safe control frameworks, we directly close the perception-action loop and demonstrate the full capability of safe control in combination with real-time VI-SLAM without any external infrastructure or prior knowledge of the environment. We verify the efficacy of the perceptive safety filter in real-time MAV experiments using exclusively onboard sensing and computation and show that the teleoperated MAV is able to safely navigate through unknown environments despite arbitrary inputs sent by the teleoperator.

关键词： Teleoperators State feedback Three-dimensional displays Simultaneous localization and mapping Navigation Real-time systems Time measurement

OculusGraphy: Description and Time Domain Analysis of Full-Field Electroretinograms Database

学校读者我要写书评

暂无评论

OculusGraphy: Description and Time Domain Analysis of Full-F...

Ural Symposium on Biomedical Engineering, Radioelectronics and information Technology (USBEREIT)

作者： Faisal B. Albasu Subhankar Dey Anton Yu. Dolganov Oussama EL Hamzaoui Wisam M. Mustafa Aleksei E. Zhdanov Engineering School of Information Technologies Telecommunications and Control Systems Ural Federal University Yekaterinburg Russia Machine Learning and Data Analytics Lab University of Erlangen-Nuremberg Erlangen Germany

Electroretinography is an established electrophysiological technique for clinical investigation of visual health function. The test involves the non-invasive assessment of the retinal function using light stimulation. This study presents a comprehensive database of five types of signals, including Scotopic 2.0 ERG response, Photopic 2.0 ERG response, Maximum 2.0 ERG response, Photopic 2.0 EGR Flicker response and Scotopic 2.0 ERG Oscillatory Potentials. The database includes 1975 signals from 323 patients, consisting of both adults and children. This study seeks to evaluate and analyze the classical parameters of three of the five signals, specifically Scotopic 2.0 ERG response, Photopic 2.0 ERG response, and Maximum 2.0 ERG response, using time-domain analysis. Correlation analysis was conducted in order to gain insights into the potential relationship between a patient's age and the classical parameters. However, the results suggest little to no correlation between these factors. The database presented in this study will serve as a useful resource for further research in the field of electrophysiology. Future research in this area could explore the use of other advanced analysis methods, such as frequency and time-frequency domain analyses, in order to gain further insights into the nuances of these signals and to better differentiate between healthy patients and those with pathologies. Additionally, further research is needed to establish norms for pediatric signals in order to better understand retinal health in children.

关键词：

Anomaly Heterogeneity learning for Open-set Supervised Anomaly Detection

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhu, Jiawen Ding, Choubo Tian, Yu Pang, Guansong School of Computing and Information Systems Singapore Management University Singapore Australian Institute for Machine Learning University of Adelaide Australia Harvard Ophthalmology AI Lab Harvard University United States

Open-set supervised anomaly detection (OSAD) - a recently emerging anomaly detection area - aims at utilizing a few samples of anomaly classes seen during training to detect unseen anomalies (i.e., samples from open-set anomaly classes), while effectively identifying the seen anomalies. Benefiting from the prior knowledge illustrated by the seen anomalies, current OSAD methods can often largely reduce false positive errors. However, these methods are trained in a closed-set setting and treat the anomaly examples as from a homogeneous distribution, rendering them less effective in generalizing to unseen anomalies that can be drawn from any distribution. This paper proposes to learn heterogeneous anomaly distributions using the limited anomaly examples to address this issue. To this end, we introduce a novel approach, namely Anomaly Heterogeneity learning (AHL), that simulates a diverse set of heterogeneous anomaly distributions and then utilizes them to learn a unified heterogeneous abnormality model in surrogate open-set environments. Further, AHL is a generic framework that existing OSAD models can plug and play for enhancing their abnormality modeling. Extensive experiments on nine real-world anomaly detection datasets show that AHL can 1) substantially enhance different state-of-the-art OSAD models in detecting seen and unseen anomalies, and 2) effectively generalize to unseen anomalies in new domains. Code is available at https://***/mala-lab/AHL. Copyright © 2023, The Authors. All rights reserved.

关键词： Anomaly detection

Reproducibility and Geometric Intrinsic Dimensionality: An Investigation on Graph Neural Network Research

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Hille, Tobias Stubbemann, Maximilian Hanika, Tom Knowledge & Data Engineering Group University of Kassel Kassel Germany Interdisciplinary Research Center for Information System Design University of Kassel Kassel Germany Information Systems and Machine Learning Lab University of Hildesheim Hildesheim Germany Institute of Computer Science University of Hildesheim Hildesheim Germany

Difficulties in replication and reproducibility of empirical evidences in machine learning research have become a prominent topic in recent years. Ensuring that machine learning research results are sound and reliable requires reproducibility, which verifies the reliability of research findings using the same code and data. This promotes open and accessible research, robust experimental workflows, and the rapid integration of new findings. Evaluating the degree to which research publications support these different aspects of reproducibility is one goal of the present work. For this we introduce an ontology of reproducibility in machine learning and apply it to methods for graph neural networks. Building on these efforts we turn towards another critical challenge in machine learning, namely the curse of dimensionality, which poses challenges in data collection, representation, and analysis, making it harder to find representative data and impeding the training and inference processes. Using the closely linked concept of geometric intrinsic dimension we investigate to which extend the used machine learning models are influenced by the intrinsic dimension of the data sets they are trained *** Codes 68T01 68T07 68T09 51F99 Copyright © 2024, The Authors. All rights reserved.

关键词： Graph neural networks