Deep Curriculum Learning in Task Space for Multi Class Based Mammography Diagnosis

Deeplearning techniques have succeeded in reaching near-human performance in anumber of tasks . Mammography is used as a standard screening procedure for the potential patients of breast cancer . We propose an end-to-endCurriculum Learning (CL) strategy in task space for classifying the three categories of Full-Field Digital Mammography (FFDM), namely Malignant,Negative, and False recall .…

Towards Reducing Aleatoric Uncertainty for Medical Imaging Tasks

In safety-critical applications like medical diagnosis, certainty associated with a model’s prediction is just as important as its accuracy . Uncertainty inpredictions can be attributed to noise or randomness in data (aleatoric) and incorrect model inferences (epistemic) While model uncertainty can be reduced with more data or bigger models, aleatoric uncertainty is more intricate .…

HCV Hierarchy Consistency Verification for Incremental Implicitly Refined Classification

Hierarchy-ConsistencyVerification (HCV) is an enhancement to existing continual learning methods . Our method incrementally discovers the hierarchical relations between classes . We then show how this knowledge can be exploited during both training andference . Code is available inhttps://://://github.com/wangkai930418/HCV_IIRC and HCV_Hierarchy-ConferenceVerification is available to download in dro dro droplets of code .…

Topic Guided Abstractive Multi Document Summarization

A critical point of multi-document summarization (MDS) is to learn therelations among various documents . In this paper, we propose a novelabstractive MDS model, in which we represent multiple documents as aheterogeneous graph . We adopt amulti-task learning strategy to jointly train the topic and summarization module, allowing the promotion of each other .…

MOS A Low Latency and Lightweight Framework for Face Detection Landmark Localization and Head Pose Estimation

Dynamic facerecognition (DFR) in wild has received much attention in recent years . Facedetection and head pose estimation are two important steps for DFR . The proposed method achieves the state-of-the-art performance in low computational resources . Another challenge is that robots often use low computational units like ARM based computing core and we often need to use lightweightnetworks instead of the heavy ones, which lead to performance drop especially for small and hard faces .…

A Real Time Energy and Cost Efficient Vehicle Route Assignment Neural Recommender System

This paper presents a neural network recommender system algorithm forassigning vehicles to routes based on energy and cost criteria . The new system has been deployed and integrated into the POLARISTransportation System Simulation Tool for use in research conducted by the Department of Energy’s Systems and Modeling for Accelerated Research inTransportation (SMART) Mobility Consortium .…

PlaneRecNet Multi Task Learning with Cross Task Consistency for Piece Wise Plane Detection and Reconstruction from a Single RGB Image

Piece-wise 3D planar reconstruction provides holistic scene understanding ofman-made environments, especially for indoor scenarios . Most recent approaches focused on improving the segmentation and reconstruction results by introducing advanced network architectures but overlooked the dual characteristics ofpiece-wise planes as objects and geometric models .…

LOA Logical Optimal Actions for Text based Interaction Games

Logical Optimal Actions (LOA) is an action decision architecture ofreinforcement learning applications with a neuro-symbolic framework . LOA is a combination of neural network and symbolic knowledge acquisition approach for natural language interaction games . The demonstration for LOA experimentsconsists of a web-based interactive platform for text-based games and visualization for acquired knowledge for improving interpretability for trained rules .…

Neuro Symbolic Reinforcement Learning with First Order Logic

Deep reinforcement learning (RL) methods often require many trials before convergence, and no direct interpretability of trained policies is provided . We propose a novel RL method for text-based games with a recent neuro-symbolicframework called Logical Neural Network . The method is first toextract first-order logical facts from text observation and external wordmeaning network (ConceptNet) then train a policy in the network with directlyinterpretable logical operators .…

LOA Logical Optimal Actions for Text based Interaction Games

Logical Optimal Actions (LOA) is an action decision architecture ofreinforcement learning applications with a neuro-symbolic framework . LOA is a combination of neural network and symbolic knowledge acquisition approach for natural language interaction games . The demonstration for LOA experimentsconsists of a web-based interactive platform for text-based games and visualization for acquired knowledge for improving interpretability for trained rules .…

Generalized Out of Distribution Detection A Survey

Out-of-distribution (OOD) detection is critical to ensuring the reliabilityand safety of machine learning systems . Other problems related to OOD detection include anomaly detection (AD), novelty detection (ND), open set recognition (OSR), and outlier detection (OD) Despite having different definitions and settings, these problems often confuse readers and practitioners, some existing studies misuse terms .…

Dual Encoding U Net for Spatio Temporal Domain Shift Frame Prediction

The landscape of city-wide mobility behaviour has altered significantly over the past 18 months . The ability to make accurate and reliable predictions on such behaviour has likewise changed drastically . This paper seeks to address this question by introducing an approach for traffic frame prediction using a lightweight dual-Encoding U-Net built using only 12 Convolutional layers .…

DAIR Data Augmented Invariant Regularization

Deep learning through empirical risk minimization (ERM) has succeeded at achieving human-level performance at a variety of complex tasks . ERM generalizes poorly to distribution shift, partly explained by overfitting to spurious features such as background in images or named entities in natural language .…

SecureBoost A High Performance Gradient Boosting Tree Framework for Large Scale Vertical Federated Learning

Gradient boosting decision tree is a widely used ensemble algorithm used in cross-silo privacy-preserving modeling . SecureBoost+ integrates several ciphertext calculationoptimizations and engineering optimizations . It makes effective andefficient large-scale vertical federated learning possible . The experimental resultsdemonstrate that Secureboost+ has significant performance improvements on largeand high-dimensional data sets compared to SecureBoost+.…

PipAttack Poisoning Federated Recommender Systems forManipulating Item Promotion

Due to growing privacy concerns, decentralization emerges rapidly in personalized services, especially recommendation . centralized models are vulnerable to poisoning attacks, compromising their integrity . In the context of recommender systems, a typical goal of suchpoisoning attacks is to promote the adversary’s target items by interfering with the training dataset and/or process .…

DAIR Data Augmented Invariant Regularization

Deep learning through empirical risk minimization (ERM) has succeeded at achieving human-level performance at a variety of complex tasks . ERM generalizes poorly to distribution shift, partly explained by overfitting to spurious features such as background in images or named entities in natural language .…

2020 CATARACTS Semantic Segmentation Challenge

The 2020 CATARACTS SemanticSegmentation Challenge was a sub-challenge of the 2020 MICCAI EndoscopicVision (EndoVis) Challenge . It presented three sub-tasks to assess participatingsolutions on anatomical structure and instrument segmentation . Their performance was assessed on a hidden test set of 531 images from 10 videos of the CATarACTS test set .…

Neuro Symbolic Reinforcement Learning with First Order Logic

Deep reinforcement learning (RL) methods often require many trials before convergence, and no direct interpretability of trained policies is provided . We propose a novel RL method for text-based games with a recent neuro-symbolicframework called Logical Neural Network . The method is first toextract first-order logical facts from text observation and external wordmeaning network (ConceptNet) then train a policy in the network with directlyinterpretable logical operators .…

Vis TOP Visual Transformer Overlay Processor

In recent years, Transformer has achieved good results in Natural LanguageProcessing (NLP) and has also started to expand into Computer Vision (CV) We propose Vis-TOP (Visual Transformer OverlayProcessor), an overlay processor for various visual Transformer models . Compared to the existing Transformer accelerators, our throughput perDSP is between 2.2x and 11.7x higher than others .…

Dual Encoding U Net for Spatio Temporal Domain Shift Frame Prediction

The landscape of city-wide mobility behaviour has altered significantly over the past 18 months . The ability to make accurate and reliable predictions on such behaviour has likewise changed drastically . This paper seeks to address this question by introducing an approach for traffic frame prediction using a lightweight dual-Encoding U-Net built using only 12 Convolutional layers .…

Transfer beyond the Field of View Dense Panoramic Semantic Segmentation via Unsupervised Domain Adaptation

Modern semantic segmentation approaches rely heavily on annotated training data which is rarely available for panoramic images . We introduce P2PDA – a generic framework for Pinhole-to-Panoramicsemantic segmentation . The framework uses uncertainty-aware adaptation using confidence values regulated on-the-fly through attention heads with discrepant predictions .…

Super resolution of multiphase materials by combining complementary 2D and 3D image data using generative adversarial networks

Modelling the impact of a material’s mesostructure on device levelperformance typically requires access to 3D image data containing all therelevant information to define the geometry of the simulation domain . In this paper, we present a method for combining information from pairs of distinct but complementary imaging techniques in order to accurately reconstruct the desired multi-phase, high resolution,representative, 3D images .…

Generative Adversarial Graph Convolutional Networks for Human Action Synthesis

Kinetic-GAN is a novel architecture that leverages the benefits of GenerativeAdversarial Networks and Graph Convolutional Networks to synthesise thekinetics of the human body . The proposed adversarial architecture can condition up to 120 different actions over local and global body movements while improving sample quality and diversity through latent space disentanglement andstochastic variations .…

Can Q learning solve Multi Armed Bantids

When a reinforcement learning (RL) method has to decide between severaloptional policies by solely looking at the received reward, it has toimplicitly optimize a Multi-Armed-Bandit (MAB) problem . We claim that the surprising answer is no. In our experiments we show that in somesituations they fail to solve a basic MAB problem .…

The Effect of Wearing a Face Mask on Face Image Quality

Wearing mouth-and-nose protection has been made a mandate in many places, to prevent the spread of the COVID-19 virus . However, face masks affect the performance of face recognition, since a large area of the face is covered . This work studies, for the first time, the effect ofwearing a face mask on face image quality by utilising state-of-the-art faceimage quality assessment methods of different natures .…