Artificial Neural Networks and Machine Learning – ICANN 2022: 31st International Conference on Artificial Neural Networks, Bristol, UK, September 6–9, ... (Lecture Notes in Computer Science, 13531)

The 4-volumes set of LNCS 13529, 13530, 13531, and 13532 constitutes the proceedings of the 31st International Conference on Artificial Neural Networks, ICANN 2022, held in Bristol, UK, in September 2022. The total of 255 full papers presented in these proceedings was carefully reviewed and selected from 561 submissions. ICANN 2022 is a dual-track conference featuring tracks in brain inspired computing and machine learning and artificial neural networks, with strong cross-disciplinary interactions and applications. Preface Organization Contents- Part III Adaptive Channel Encoding Transformer for Point Cloud Analysis 1 Introduction 2 Related Work 2.1 Point-Based Method 2.2 Transformer-Based Method 3 Method 3.1 Tce 3.2 Network Architecture 4 Experiments 4.1 Classification on ModelNet40 4.2 Part Segmentation on ShapeNet 4.3 Classification on ScanObjectNN 4.4 Ablation Studies 4.5 Robustness Experiments 5 Conclusion References ARB U-Net: An Improved Neural Network for Suprapatellar Bursa Effusion Ultrasound Image Segmentation 1 Introduction 2 ARB U-Net Network Structure 2.1 Encoder 2.2 Decoder 2.3 Loss Function 3 Result 3.1 Experimental Environment and Dataset Annotation 3.2 Evaluation Index 3.3 Quantitative and Qualitative Analysis 4 Conclusion References BPGG: Bidirectional Prototype Generation and Guidance Network for Few-Shot Anomaly Localization 1 Introduction 2 Related Work 2.1 Anomaly Localization 2.2 Few-Shot Learning 2.3 Few-Shot Segmentation 3 Method 3.1 Problem Definition 3.2 Method Overview 3.3 Prototype Generation and Guidance 3.4 Original Forward Branch 3.5 Adaptive Reverse Branch 4 Experiments 4.1 Experiment Setup 4.2 Qualitative Results 4.3 Ablation Study 5 Conclusions References CoPrGAN: Image-to-Image Translation via Content Preservation 1 Introduction 2 Related Work 3 Approach 3.1 Model Overview 3.2 Dynamic Paths 3.3 Training Strategy 4 Experiments 4.1 Datasets 4.2 Baselines 4.3 Quality Comparison 4.4 Ablation Study 4.5 Discussions 5 Conclusion References Cross Domain Evaluation of Text Detection Models 1 Introduction 2 Related Work 2.1 Character-Based Detectors 2.2 Word-Based Detectors 2.3 Line-Based Detectors 2.4 Segmentation-Based Detectors 3 Models 3.1 EAST 3.2 CRAFT 3.3 Tesseract 3.4 Outputs Ensemble 4 Experiment 4.1 Datasets 4.2 Experimental Set-Up 5 Results and Discussion 6 Conclusion References Cross-Domain Learning for Reference-Based Sketch Colorization with Structural and Colorific Strategy 1 Introduction 2 Related Work 2.1 Conditional Sketch Colorization 2.2 Reference-Based Colorization 3 Proposed Method 3.1 Domain Alignment Network 3.2 Coarse-to-Fine Generator 3.3 Structural and Colorific Strategy 3.4 Loss for Reference-Based Sketch Colorization 4 Experiments 4.1 Implementation 4.2 Datasets 4.3 Qualitative Comparison 4.4 Ablation Study 4.5 User Research 5 Conclusion References Data Augmented Dual-Attention Interactive Image Classification Network 1 Introduction 2 Related Work 2.1 Method Based on Strong Supervision 2.2 Method Based on Weak Supervision 3 Method 3.1 Data Augmentation 3.2 Dual Attention Network 3.3 Channel Interaction and Local Feature Fusion 4 Experimental Results and Analysis 4.1 Datasets and Implementation Details 4.2 Experimental Results on Three Fine-Grained Data Sets 4.3 Data Augmentation and Dual Attention Visualization 4.4 Ablation Experiments 5 Conclusions References Deep Dictionary Pair Learning for SAR Image Classification 1 Introduction 2 Related Work 2.1 Discriminative Dictionary Learning 2.2 Deep Convolutional Neural Network 3 Proposed Method 3.1 Projective Dictionary Pair Learning 3.2 Dictionary Learning Layers 3.3 Network Architecture 3.4 Training and Inference 4 Experiments 4.1 DataSet 4.2 Setting 4.3 Results 5 Discussion 5.1 Ablation Study 5.2 Parameter Comparison Study 6 Conclusion References Deepfake Video Detection Exploiting Binocular Synchronization 1 Introduction 2 Related Work 3 Proposed Method 3.1 Preprocessing 3.2 Architecture 4 Experiment and Analysis 4.1 Datasets 4.2 Implementation Details 4.3 Feature Effectiveness 4.4 In-Dataset Evaluation 4.5 Cross-Dataset Evaluation 4.6 Ablation Study 4.7 Limitations 5 Conclusion References Dep-ViT: Uncertainty Suppression Model Based on Facial Expression Recognition in Depression Patients 1 Introduction 2 VFEM 2.1 Expression Collection 2.2 Emoticon Scoring 2.3 Facial Feature Extraction and Analysis 3 Dep-ViT 3.1 Encoder 3.2 SE and Self-attention Layer 3.3 Rank Regularization Based on KL Divergence and Relabeling 3.4 Loss Function Based on Manual Labeling 4 Experiment and Result Analysis 4.1 Parameter Settings 4.2 Experimental Results 4.3 Ablation Experiment 5 Limitation 6 Conclusion References Ensemble of One-Class Classifiers Based on Multi-level Hidden Representations Abstracted from Convolutional Autoencoder for Anomaly Detection 1 Introduction 2 Related Work 2.1 OCSVMs 2.2 Hybrid Approach 3 Ensemble of One-Class Classifiers 3.1 Extracting Different Levels of Image Semantic Features 3.2 Building Multiple Base Classifiers with Extracted Features 3.3 Classifier Fusion for Image Anomaly Evaluation 4 Experiments 4.1 Datasets and Setup 4.2 Baseline Methods 4.3 Model Configuration 4.4 Performance Evaluation Metric 4.5 Results and Analysis 5 Conclusion and Future Work References Images Structure Reconstruction from fMRI by Unsupervised Learning Based on VAE 1 Introduction 2 Method 2.1 Basic Model and Loss Function 2.2 Overview of the Proposed Framework 3 Experimental Results 3.1 Datasets and Evalution 3.2 Comparison of Images Reconstruction Performance with Others 3.3 Ablation Experiment 4 Conclusion References Inter-subtask Consistent Representation Learning for Visual Commonsense Reasoning 1 Introduction 2 Related Work 2.1 Visual Commonsense Reasoning 2.2 Siamese Network 2.3 Contrastive Learning 3 Proposed Approach 3.1 Joint Learning Framework 3.2 Feature Extraction 3.3 Multi-level Contrastive Learning 3.4 Classification and Loss 4 Experiments 4.1 Datasets and Implementation Details 4.2 Performance Comparison 4.3 Ablation Studies 5 Conclusion References InvisibiliTee: Angle-Agnostic Cloaking from Person-Tracking Systems with a Tee 1 Introduction 1.1 User Study 2 Literature Review 3 Method 3.1 Overview 3.2 Attack and Geometric Constraint Loss Functions 3.3 Geometric Warp and Masking 4 Attacks in the Digital World 4.1 Dataset and Experiment Setup 4.2 Experimental Results 4.3 Case Studies of Digital Attacks 5 Attacks in the Physical World 5.1 Additional Discussion 6 Conclusion References Makeup Transfer Based on Generative Adversarial Network for Large Angle Spatial Misalignment 1 Introduction 2 Related Work 2.1 Facial Makeup Transfer 2.2 Style Transfer 3 Related Work 3.1 Formulation 3.2 Framework 3.3 Neural Head Reenactment Module 3.4 Full Objective 4 Experiments 4.1 Implementation Details 4.2 Comparisons 4.3 Ablation Studies 4.4 Controllable Makeup Transfer 4.5 Partial Makeup Transfer 4.6 Makeup Remove 5 Conclusion References Making Images Resilient to Adversarial Example Attacks 1 Introduction 2 Adversary-Proof Examples 2.1 Baseline: R-PGD 2.2 Advanced: ZigZag 3 Empirical Study 3.1 Experiment Settings 3.2 R-PGD Against FGSM 3.3 R-PGD Against PGD 3.4 ZigZag Against FGSM 3.5 ZigZag Against PGD 3.6 ZigZag Against CW 3.7 Transferability Evaluation 4 Related Work 5 Concluding Remarks References Multi-Class Lane Semantic Segmentation of Expressway Dataset Based on Aerial View 1 Introduction 2 Related Work 2.1 Lane Detection Datasets 2.2 Semantic Segmentation Models Based on DCNNs 2.3 Hausdorff Distance Loss 2.4 Conditional Random Fields 3 Multi-class Lane Semantic Segmentation 3.1 Expressway Dataset Based on Aerial View 3.2 DeepLab-ERFC 3.3 Update Strategy 4 Experiment 4.1 Comparison Experiment 4.2 Evaluation on Expressway Dataset 5 Conclusion References Mutil-level Local Alignment and Semantic Matching Network for Image-Text Retrieval 1 Introduction 2 Related Work 3 Method 3.1 Generic Representation Extraction 3.2 Local Region-Word Alignment 3.3 Multi-level Semantic Matching 3.4 Loss Function 4 Experiments 4.1 Dataset and Evaluation Metric 4.2 Implementation Details 4.3 Comparison with State-of-the-Art Methods 4.4 Ablation Study and Analysis 4.5 Visualization of Retrieval Results 5 Conclusion References NAS4FBP: Facial Beauty Prediction Based on Neural Architecture Search 1 Introduction 2 Related Work 2.1 Facial Beauty Prediction Based on Deep Learning 2.2 Neural Architecture Search 3 Method 3.1 Align-Crop 3.2 NAS for an FBP Backbone 3.3 Non-local Spatial Attention Module 3.4 Multi-task Learning Scheme with HBLoss 4 Experimental Results 4.1 Experimental Setup 4.2 Experiments of Applying NAS to FBP 4.3 Comparison with the Related State-of-the-Art Models 4.4 Ablation and Analysis 5 Conclusion References Object Detector with Recursive Feature Pyramid and Key Content-Only Attention 1 Introduction 2 Related Works 3 RecursiveFeaturePyramid 4 Key Content-Only Attention 5 Experiments 5.1 Experimental Details 5.2 Ablation Studies 5.3 Main Results 6 Conclusion References O-LGMD: An Opponent Colour LGMD-Based Model for Collision Detection with Thermal Images at Night 1 Introduction 2 Related Work 2.1 Colour Models 2.2 LGMD-Based Collision Detection Model 3 Proposed Method 3.1 The oRGB Transformation 3.2 The LGMD-Based Neural Network 4 Experiments and Discussion 4.1 Experimental Setup 4.2 Colour Discrimination Test 4.3 Pseudo-colour Thermal Clips Test 4.4 Discussion 5 Conclusion References Semantic Diversity Image Translation Based on Deep Feature Difference and Attention Mechanism 1 Introduction 2 Related Work 3 Our Model 3.1 Framework 3.2 Deep Feature Diversity 3.3 Attention Mechanism 3.4 Objective Function 4 Experiment 4.1 Experiment Setting 4.2 Experiment Result 4.3 Ablation Study 5 Conclusion References Spatial Foreground Bigraph Matching for Generalizable Person Re-identification 1 Introduction 2 Related Work 2.1 Generalizable Person Re-identification 2.2 Partial Person Re-identification 3 The Proposed Approach 3.1 Contextual Foreground Region Extraction 3.2 Feature Bigraph Module and Part-Part Match Strategy 3.3 Final Distance Definition 4 Experiments 4.1 Experimental Settings 4.2 Results 4.3 Ablation Study and Visualization 5 Conclusion References TSN-CA: A Two-Stage Network with Channel Attention for Low-Light Image Enhancement 1 Introduction 2 Related Work 2.1 Traditional Methods 2.2 Deep Learning-Based Methods 3 Motivation and Analysis 4 Methodology 4.1 Channel Attention for Image Restoration 5 Experiments 5.1 Implementation 5.2 Ablation Study 5.3 Quantitative Comparison 5.4 Qualitative Comparison References Two-Stream Interactive Memory Network for Video Facial Expression Recognition 1 Introduction 2 Related Work 2.1 Memory Network 2.2 Attention Mechanism 3 Our Proposed Framework 3.1 Feature Extraction Network 3.2 Two-Stream Interactive Memory Network 4 Experiment 4.1 Datasets 4.2 Comparison of Different Computational Layers (Hops) 4.3 Performance Analysis of TM-CSA 4.4 Model Component Analysis 4.5 Visualization of Attention Map 5 Conclusion References Utilize Spatial Prior in Ground Truth: Spatial-Enhanced Loss for Semantic Segmentation 1 Introduction 2 Related Work 2.1 Focus on Hard Pixels 2.2 Class Imbalance 2.3 Dice Loss 3 Method 3.1 Separation of the Body and Edge Region 3.2 Rescale Based on Regions 4 Experiments 4.1 Implement Details 4.2 Analysis of Threshold Distance d 4.3 Analysis of Region Weights 4.4 Comparison with Other Methods 5 Conclusion References Weighted Pooling from Salient Regions for Place Recognition 1 Introduction 2 Related Work 3 Method 3.1 Feature Extraction 3.2 Salient Regions Detection 3.3 Feature Fusion by Weight Mask 3.4 D. Learning with Triple Loss 4 Experiment 4.1 Datasets and Evaluation Methodology 4.2 Implementation Details 4.3 Experimental Results 5 Conclusion References Efficient LSTM Training with Eligibility Traces 1 Introduction 2 E-Prop 2.1 Extensions to LSTM Based E-Prop 2.2 E-Prop in the Context of Reinforcement Learning 3 Experiments and Results 3.1 Classification of Handwritten Digits 3.2 Solving Another Temporal Credit Assignment Problem 4 Solving Easy RL Environments with DRQN 4.1 Environment 5 Discussion 6 Conclusion References Improving Stylized Image Captioning with Better Use of Transformer 1 Introduction 2 Related Work 3 Methodology 3.1 Base Model: CapTransformer 3.2 Multi-step Fused Attention Network 3.3 Gated Decoder Model 3.4 Overall Learning Strategy 4 Experiments 4.1 Dataset 4.2 Implementation Details 4.3 Results 5 Human Evaluation 5.1 Ablation Study 5.2 Case Study 6 Conclusion References Investigating Current-Based and Gating Approaches for Accurate and Energy-Efficient Spiking Recurrent Neural Networks 1 Introduction 2 Related Work 2.1 Leaky Integrate-and-Fire and Current-Based Models 2.2 Gated Recurrent Networks 3 SpikGRU: A Spiking Gated Recurrent Unit 4 Experiments 4.1 Methods 4.2 Results and Discussion 5 Energy-Efficiency 6 Conclusion and Perspectives References Research on USV Path Planning Method Based on Improved Option-Critical Algorithm 1 Introduction 2 Materials and Methods 2.1 Option-Critic Algorithm 2.2 Option-Critc Framework Based on RNN 3 Algorithm 3.1 Environment and Reward Design 3.2 Algorithm 4 Simulation Experiment 5 Conclusion References Bi-level Optimization Method for Automatic Reward Shaping of Reinforcement Learning 1 Introduction 2 Algorithm Design 2.1 The Theoretical Basis and Problem Description of Reward Shaping 2.2 Algorithm Design Based on AutoMLzero 3 Improvement of Search Efficiency 4 Experiments 4.1 Experimental Setup 4.2 Analysis of Results 5 Conclusion References Category-Guided Localization Network for Visual Sound Source Separation 1 Introduction 2 Method 2.1 Overview 2.2 Image-Attention Branch 2.3 Video-Attention Branch and Separation 3 Experiment 3.1 Datasets 3.2 Implementation Details 4 Results and Discussion 4.1 Separation Results 4.2 Ablation Study 4.3 Qualitatively Analysis 5 Conclusion References Gaussian Mixture Model-Based Registration Network for Point Clouds with Partial Overlap 1 Introduction 2 Related Works 3 IT2GMM-Net 3.1 Feature Extractor 3.2 SegNet and ClaNet Module 3.3 GMM and T Calculation Module 3.4 Loss 4 Experiments 4.1 ModelNet40 4.2 Ablation Experiments and Computational Efficiency 5 Conclusion References More Diverse Training, Better Compositionality! Evidence from Multimodal Language Learning 1 Introduction 2 Related Work 3 Multimodal Dataset 4 Model 5 Training Setup 6 Results 7 Discussion 8 Conclusion References MT-TCCT: Multi-task Learning for Multimodal Emotion Recognition 1 Introduction 2 Related Work 3 Approach 3.1 Modality Feature Learning 3.2 Emotion Recognition 3.3 Prediction Layer 3.4 Optimization Objectives 4 Experiments 4.1 Datasets 4.2 Evaluation Metrics 4.3 Basic Settings 4.4 Baselines 5 Results and Analysis 5.1 Quantitative Results 5.2 Ablation Study 6 Conclusion References NeoSLAM: Neural Object SLAM for Loop Closure and Navigation 1 Introduction 2 Related Works 3 Description of the NeoSLAM System 3.1 Object Detection and Description for Scene Understanding with Appearance 3.2 Keyframe Detection and Loop Closure Detection 3.3 Training the Grid Cell (GCN) and the Head Direction Cell Networks (HDN) 3.4 Modeling the Firing Fields of the Object Vector Cells 3.5 Learning Associations Between Keyframe Templates and GCN 3.6 Object Experience Map Creation 4 Discussion and Results 4.1 Training the OVCN 4.2 Loop Closure Detection 4.3 Object Experience Map Creation and Update 5 Conclusion References Trajectory-Based Mobile Game Bots Detection with Gaussian Mixture Model 1 Introduction 2 Related Work 3 Preliminary 3.1 Problem Definition 4 Methodology 4.1 Preprocess 4.2 Multi-view GM-TVAE 5 Experiment 5.1 Experimental Settings 5.2 Experimental Results 5.3 Ablation Study 5.4 Hyper-parameter Studies 6 Conclusion References Unsupervised Domain Adaptation Using Temporal Association for Segmentation and Its Application to C. elegans Time-Lapse Images 1 Introduction 2 Related Work 3 Proposed Method 3.1 Problem Definition in Video-Based UDA 3.2 Overall Idea and Network 3.3 Training Steps 4 Experiments 4.1 Datasets 4.2 Experimental Conditions 4.3 The Case with the Same Time Intervals in the Domains 4.4 The Case with Different Time Intervals Between the Domains 5 Conclusion References Unsupervised Multi-view Multi-person 3D Pose Estimation Using Reprojection Error 1 Introduction 2 Related Work 3 UMVpose 3.1 Reprojection Error 3.2 Matching Process 3.3 Loss Function 3.4 Optimizer 4 Experimental Setup 4.1 Dataset 4.2 Metrics 5 Results and Discussions 6 Conclusion References Visual Relation-Aware Unsupervised Video Captioning 1 Introduction 2 Related 2.1 Supervised Video Captioning 2.2 Unsupervised Captioning 3 Method 3.1 Pseudo Captions Retrieve 3.2 Basic Video Captioning Model 3.3 Visual Relation-Aware Module 4 Experiments 4.1 Datasets and Metrics 4.2 Implementation Details 4.3 Quantitative Results 4.4 Qualitative Results 5 Conclusion References A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model 1 Introduction 2 Izhikevich's Spiking Neuron Model 3 Folding Architecture for a Single Neuron 4 FPGA Implementation and Simulation 5 Results and Discussion 6 Conclusion References A Spiking Neural Network Based on Neural Manifold for Augmenting Intracortical Brain-Computer Interface Data 1 Introduction 2 Background and Related Work 2.1 Spiking Neuron 2.2 Neural Manifold 2.3 Neural Decoding 2.4 Related Work 3 Method 3.1 Overview Architecture 3.2 Surrogate Gradient Descent 3.3 Generating Spike Trains Dataset 4 Experiments 4.1 Data Processing and Analysis 4.2 Experimental Setting 4.3 Results 5 Conclusion Reference s An Improved Lightweight YOLOv5 Model Based on Attention Mechanism for Face Mask Detection 1 Introduction 2 Related Works 2.1 Object Detection 2.2 Face Mask Detection 3 Methodology 3.1 Data Processing 3.2 Network Architecture 4 Experiments and Results 4.1 Dataset 4.2 Evaluation Metrics 4.3 Implementation Details 4.4 Ablation Study 4.5 Comparative Results with Other Models on AIZOO 5 Conclusion References Analytical Comparison Between the Pattern Classifiers Based upon a Multilayered Perceptron and Probabilistic Neural Network in Parallel Implementation 1 Introduction 2 Implementing an MLP-NN and PNN into Parallel Computing Environment 2.1 Feed-Forward Computation of an MLP-NN in Parallel 2.2 Feed-Forward Computation of a PNN in Parallel 2.3 Comparison of the Reference Mode Between an MLP-NN and PNN 2.4 Reduction in the Number of the RBFs in a PNN via k-Means Clustering 2.5 K-Means Clustering Algorithm Operated in a Parallel Environment 2.6 Comparing the Training Mode of an MLP-NN with that of a Compact PNN Obtained Using the k-Means Clustering 3 Empirical Justification of the Criteria (8) and (13) 4 Conclusion References Architecture-Agnostic Time-Step Boosting: A Case Study in Short-Term Load Forecasting 1 Introduction 2 Related Works 3 Time-Step Boosting 4 Experimental Evaluation 4.1 Data Sets and Preprocessing 4.2 Experiments and Results 5 Conclusions and Future Work References ASTra: A Novel Algorithm-Level Approach to Imbalanced Classification 1 Introduction 2 Background and Related Work 2.1 The Confusion Matrix and Its Associated Metrics 2.2 Related Work 3 Proposed Methods 3.1 The ASTra (Asymmetric Sigmoid Transfer) Function 3.2 The GMN Loss Function 4 Data, Experimental Process, and Performance Measurement 4.1 Data 4.2 Details of the Experimental Process 4.3 Performance Measurement 5 Results 6 Discussion References Attention Awareness Multiple Instance Neural Network 1 Introduction 2 Related Work 3 Preliminary 4 Methodology 4.1 Network Overview 4.2 Instance-Level Classifier 4.3 Attention Based MIL Pooling 4.4 Bag-Level Classification Layer 5 Experiment and Analysis 5.1 Classic MIL Datasets 5.2 Aerial Image Dataset 5.3 Medical Image Dataset 5.4 MIL Document Dataset 6 Conclusion References Context Reasoning Attention Network: Generating Plausible Distractors for Multi-choice Questions 1 Introduction 2 Problem Definition and Related Work 3 Methodology 3.1 Encoding 3.2 Decoding 3.3 Training Details 4 Experiments 4.1 Dataset and Implementation Details 4.2 Baselines and Evaluation Metrics 4.3 Main Results 4.4 Ablation Study 4.5 Human Evaluation 4.6 Case Study 5 Conclusions References DensEMANN + Sparsification: Experiments for Further Shrinking Already Small Automatically Generated DenseNet 1 Introduction 2 On “Self-structuring” and Related Research 2.1 Growing and Constructive Algorithms: The Incremental Approach 2.2 Pruning and Sparsification Algorithms: The Decremental Approach 3 Methodology 3.1 Tested Sparsification Methods 3.2 Experimental Design 4 Results and Observations 4.1 First Round: Trying Various Sparsification Algorithms on Human-Designed DenseNet 4.2 Second Round: Further Sparsification Experiments, and Comparison with DensEMANN 4.3 Third Round: Applying Sparsification on DensEMANN-Generated Networks 5 Conclusions and Future Work References Dynamic Vision Sensor Based Gesture Recognition Using Liquid State Machine 1 Introduction 2 Background 2.1 Dynamic Vision Sensor 2.2 Liquid State Machine 2.3 STDP Learning Rule 3 Related Work 3.1 CNN for Gesture Recognition 3.2 SNN for Gesture Recognition 4 Method 4.1 Workflow for Gesture Recognition 4.2 Preprocessing 4.3 LSM Structure and Parameter Search 4.4 Real-Time Gesture Recognition System 5 Experiments 5.1 Experiment Setup 5.2 Experiment Result 6 Conclusion References Efficient and Accurate Text Detection Combining Differentiable Binarization with Semantic Segmentation 1 Introduction 2 Related Work 3 ViT-Bilateral DBNet 3.1 Overall Framework of the Integrated Network 3.2 ViT-Bilateral Network 3.3 Attention-driven Aggregation Layer 3.4 Loss Function 4 Experiments 4.1 Datasets 4.2 Implementation details 4.3 Ablation Study 4.4 Comparison 5 Conclusion References From Open Set Recognition Towards Robust Multi-class Classification 1 Introduction 2 Related Work 3 Informer 4 Evaluation Approach 5 Results 6 Conclusion References Gait Adaptation After Leg Amputation of Hexapod Walking Robot Without Sensory Feedback 1 Introduction 2 Related Work 3 Problem Specification 4 Proposed Locomotion Adaptation to Leg Amputation 4.1 Morphology Information 4.2 Gait Swing Control 4.3 ICRs Violation Detection 4.4 Phase Action Values Ordering 4.5 Swing Duration Adjustment 5 Experimental Results 6 Conclusion References Hierarchical Dynamics in Deep Echo State Networks 1 Introduction 2 Deep Echo State Networks 3 Hierarchical Dynamics and Asymptotic Stability 4 Experiments 4.1 Memory Capacity 4.2 Regression Tasks 5 Conclusions References Jacobian Ensembles Improve Robustness Trade-Offs to Adversarial Attacks 1 Introduction 2 Background 2.1 Universal Adversarial Perturbations 2.2 Model Ensembles 2.3 Jacobian Regularization 3 Bounds on UAP Effectiveness for Model Ensembles 4 Experiments with Jacobian Ensembles 4.1 Experimental Setup 4.2 Improvements with Jacobian Ensembles 4.3 Accuracy-Robustness Trade-Off 5 Conclusion References Liquid State Machine on Loihi: Memory Metric for Performance Prediction 1 Introduction 2 Background on Performance Metrics 2.1 Lyapunov Exponent 2.2 Memory Metric 3 Loihi vs Matlab 3.1 Neuronal Model Used in Loihi 3.2 Neuron Level 3.3 Reservoir Level 3.4 TI-46 Spoken Digit Recognition 4 Analysis of Memory Metric 4.1 Effect of Weight Scaling (w) 4.2 Effect of Averaging Window (win) 5 Conclusion References LogBERT-BiLSTM: Detecting Malicious Web Requests 1 Introduction 2 Related Work 3 LogBERT-BiLSTM 3.1 Preprocessing 3.2 Tokenization and Transformer-Based Classification 4 Experiments 4.1 Datasets 5 Discussion and Results 5.1 Evaluating the Performance of Models 6 Conclusion and Future Work References ML-FORMER: Forecasting by Neighborhood and Long-Range Dependencies 1 Introduction 2 Problem Formulation and Overview of the Proposed Model 2.1 Problem Formulation 2.2 Overview of the Proposed Method 3 Methodology 3.1 Time-Series Embedding 3.2 Time-Series Encoder-Decoder 3.3 Long Sequence Time-Series Loss Function 4 Evaluation 4.1 Experimental Settings 4.2 Main Results 4.3 Study of the Long Sequence Time-Series Loss Function 4.4 Ablation Study 4.5 Study of GPU Memory Footprint 5 Conclusion References Real-Time Display of Spiking Neural Activity of SIMD Hardware Using an HDMI Interface 1 Introduction 2 HEENS Architecture 3 Real-Time Display Implementation 3.1 Screen Display via HDMI 3.2 Communication with HEENS 3.3 Raster Plot 4 Experimental Results 4.1 Leaky Integrate and Fire Model 4.2 Izhikevich Model 5 Conclusion References Sailfish: A Fast Bayesian Change Point Detection Framework with Gaussian Process for Time Series 1 Introduction 2 The Design of Sailfish Framework 2.1 Problem Definition 2.2 The Overview of Sailfish Framework 2.3 Input Layer 2.4 Prediction Layer 2.5 Comparison Layer and Detection Layer 3 Experiments and Evaluation 3.1 Datasets 3.2 Baseline 3.3 Evaluation Metrics 3.4 Sensitivity Analysis 3.5 Best F1 Comparison 3.6 Comparison Table 4 Conclusion References SAM-kNN Regressor for Online Learning in Water Distribution Networks 1 Introduction 2 Related Work 3 Problem Setting 4 Model 4.1 Model Parameters 4.2 Model Adaption 4.3 Cleaning Process 4.4 Compression of the LTM 4.5 Final Model 5 Experiments 5.1 Data 5.2 Setup 5.3 Results 6 Summary and Conclusion References Spatial-Temporal Semantic Generative Adversarial Networks for Flexible Multi-step Urban Flow Prediction 1 Introduction 2 Related Works 2.1 Spatial-Temporal Urban Flow Prediction 2.2 Generative Adversarial Networks 3 Preliminaries 4 Methodology 4.1 Spatial-Temporal Semantic Encoder 4.2 Generator G 4.3 Discriminator D 4.4 The Objective Function of STS-GAN: A Weighted and Integrated Loss Function 5 Experiments 5.1 Datasets and Implementation Details 5.2 Evaluation Metrics and Baseline Methods 5.3 Experiment Results and Sensitive Study 6 Conclusion References Topic-Grained Text Representation-Based Model for Document Retrieval 1 Introduction 2 Related Work 3 TGTR 3.1 Preliminary 3.2 Model Architecture 4 Experiment Methodology 4.1 Datasets 4.2 Baseline Methods 5 Experiment Details 5.1 Implementation Details 5.2 Experiment Results 6 Analysis 6.1 A Comparison of Trade-Off Quality 6.2 Embeddings Dimension and Bytes per Dimension 7 Conclusions References Training 1-Bit Networks on a Sphere: A Geometric Approach 1 Introduction 2 Related Work 3 Preliminaries: Optimization on Riemannian Manifolds 4 Training 1-Bit DNNs on a Sphere 4.1 Projecting Filters, Layers and Entire Networks 4.2 Implementation Remarks 4.3 Relation to Spectral and Lipschitz Normalizations 5 Angle Penalization 5.1 Compatible Approaches 6 Conditioning Analysis 7 Experiments 7.1 Evaluation on CIFAR 7.2 Evaluation on ImageNet 8 Ablation Studies 9 Conclusion References A Neural Network Approach to Estimating Color Reflectance with Product Independent Models 1 Introduction 1.1 Color Matching and Color Prediction Models 2 Methodology, Analyses and Results 2.1 Dataset and Metrics 2.2 Descriptions of the Variables 2.3 ANN Product Independent Model: Implementation and Results 3 Conclusion and Future Work References Linear Self-attention Approximation via Trainable Feedforward Kernel References Author Index