CrossMinds.ai

The new homepage is released! Search for trending topics, check out weekly hype researches, follow conferences, and explore knowledge fields now!

Visit now: https://crossminds.ai

Tell us what you think about our new homepage!

Check out our new interactive research graph maps key knowledge areas and related papers for the new paper “Approximating How Single Head Attention Learns” authored by researchers Charlie Snell, Ruiqi Zhong, Dan Klein, and Jacob Steinhardt from University of California, Berkeley.



https://bit.ly/3fpeuYB

Approximating How Single Head Attention Learns - Crossminds Key knowledge areas and research papers related to the new publication “Approximating How Single Head Attention Learns” authored by Charlie Snell, Ruiqi Zhong, Dan Klein, and Jacob Steinhardt from UC ...

[ Paper] Dynamic Plane Convolutional Occupancy Networks by researchers from ETH Zürich and Max Planck Institute for Intelligent Systems.
https://bit.ly/3pdA6ID

Dynamic Plane Convolutional Occupancy Networks - Crossminds Abstract: Learning-based 3D reconstruction using implicit neural representations has shown promising progress not only at the object level but also in more complicated scenes. In this paper, we propose Dynamic Plane Convolutional Occupancy Networks, a novel implicit representation pushing further th...

[ Paper] EVET: Enhancing Visual Explanations of Deep Neural Networks Using Image Transformations by researchers from 삼성SDS (Samsung SDS).
https://bit.ly/3aSCRtJ

EVET: Enhancing Visual Explanations of Deep Neural Networks Using Image Transformations - Crossminds Abstract: Numerous interpretability methods have been developed to visually explain the behavior of complex machine learning models by estimating parts of the input image that are critical for the model's prediction. We propose a general pipeline of enhancing visual explanations using image transfor...

[ Paper] Let's Get Dirty: GAN Based Data Augmentation for Camera Lens Soiling Detection in Autonomous Driving by researchers from Valeo and Study at CTU in Prague.
https://bit.ly/3rHM2E3

Facial Emotion Recognition with Noisy Multi-task Annotations - Crossminds Abstract: Human emotions can be inferred from facial expressions. However, the annotations of facial expressions are often highly noisy in common emotion coding models, including categorical and dimensional ones. To reduce human labelling effort on multi-task labels, we introduce a new problem of fa...

[ Paper] Audio- and Gaze-driven Facial Animation of Codec Avatars by researchers from Facebook AI and Rheinische Friedrich-Wilhelms-Universität Bonn.
https://bit.ly/3oTa2SW

Audio- and Gaze-driven Facial Animation of Codec Avatars - Crossminds Abstract: Codec Avatars are a recent class of learned, photorealistic face models that accurately represent the geometry and texture of a person in 3D (i.e., for virtual reality), and are almost indis...

[ Paper] Improving Point Cloud Semantic Segmentation by Learning 3D Object Detection by researchers from Computer Vision Lab.
https://bit.ly/3tpczYD

Improving Point Cloud Semantic Segmentation by Learning 3D O - Crossminds Abstract: Point cloud semantic segmentation plays an essential role in autonomous driving, providing vital information about drivable surfaces and nearby objects that can aid higher level tasks such a...

[ Paper] Fast Fourier Intrinsic Network by researchers from Tampere University, King's College London, Study at CTU in Prague.
https://bit.ly/2LcrpQT

Fast Fourier Intrinsic Network - Crossminds Abstract: We address the problem of decomposing an image into albedo and shading. We propose the Fast Fourier Intrinsic Network, FFI-Net in short, that operates in the spectral domain, splitting the i...

[ Paper] Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows by researchers from Leibniz Universität Hannover.
https://bit.ly/3pVm1AE

Same Same But DifferNet: Semi-Supervised Defect Detection wi - Crossminds Abstract: The detection of manufacturing errors is crucial in fabrication processes to ensure product quality and safety standards. Since many defects occur very rarely and their characteristics are m...

[ Paper] Spike-Thrift: Towards Energy-Efficient Deep Spiking Neural Networks by Limiting Spiking Activity via Attention-Guided Compression by authors from USC Viterbi School of Engineering.
https://bit.ly/3af1gcQ

Spike-Thrift: Towards Energy-Efficient Deep Spiking Neural Networks by Limiting Spiking Activity via Attention-Guided Compression - Crossminds Abstract: The increasing demand for on-chip edge intelligence has motivated the exploration of algorithmic techniques and specialized hardware to reduce the computing energy of current machine learning models. In particular, deep spiking neural networks (SNNs) have gained interest because their even...

[New Paper] Robust Odometry and Mapping for Multi-LiDAR Systems with Online Extrinsic Calibration by researchers from The Hong Kong University of Science and Technology - HKUST.
https://bit.ly/3aaapmT

Robust Odometry and Mapping for Multi-LiDAR Systems with Online Extrinsic Calibration - Crossminds Abstract: Combining multiple LiDARs enables a robot to maximize its perceptual awareness of environments and obtain sufficient measurements, which is promising for simultaneous localization and mapping (SLAM). This paper proposes a system to achieve robust and simultaneous extrinsic calibration, odo...

[New Research Paper] DualSR: Zero-Shot Dual Learning for Real-World Super-Resolution by researchers from TU Eindhoven.

[ Paper] ClassMix: Segmentation-Based Data Augmentation for Semi-Supervised Learning by researchers from Chalmers University of Technology.
https://bit.ly/2MpW9OD

ClassMix: Segmentation-Based Data Augmentation for Semi-Supervised Learning - Crossminds Abstract: The state of the art in semantic segmentation is steadily increasing in performance, resulting in more precise and reliable segmentations in many different applications. However, progress is limited by the cost of generating labels for training, which sometimes requires hours of manual lab...

[New Paper] OF-VO: Reliable Navigation among Pedestrians Using Commodity Sensors by researchers from University of Maryland, College Park .
https://bit.ly/3ixeW6I

OF-VO: Reliable Navigation among Pedestrians Using Commodity Sensors - Crossminds Abstract: We present a modified velocity-obstacle (VO) algorithm that uses probabilistic partial observations of the environment to compute velocities and navigate a robot to a target. Our system uses commodity visual sensors, including a mono-camera and a 2D Lidar, to explicitly predict the velocit...

[ Paper] In-Hand Object Dynamics Inference using Tactile Fingertips by researchers from The University of Utah and NVIDIA.
https://bit.ly/3cczAIj

In-Hand Object Dynamics Inference using Tactile Fingertips - Crossminds Abstract: Having the ability to estimate an object's properties through interaction will enable robots to manipulate novel objects. Object's dynamics, specifically the friction and inertial parameters have only been estimated in a lab environment with precise and often external sensing. Could we inf...

[New Paper] Catching Out-of-Context Misinformation with Self-supervised Learning by researchers from TU München and Google AI.
https://bit.ly/3irm0lc

[TUM & Google] Catching Out-of-Context Misinformation with Self-supervised Learning - Crossminds Project: Despite the recent attention to DeepFakes and other forms of image manipulations, one of the most prevalent ways to mislead audiences is the use of unaltered images in a new but false context. To address these challenges and support fact-checkers, we propose a new method that automatically....

[ Paper] Deep Lighting Environment Map Estimation from Spherical Panoramas by researchers from Centre for Research and Technology Hellas and Universidad Politécnica de Madrid (Oficial).
https://bit.ly/3bLWndE

Deep Lighting Environment Map Estimation from Spherical Panoramas - Crossminds Estimating a scene’s lighting is a very important task when compositing synthetic content within real environments, with applications in mixed reality and post-production. In this work, we present a data-driven model that estimates an HDR lighting environment map from a single LDR monocular spheri...

[ Paper] Exertion-aware Path Generation by researchers from George Mason University and UMass Boston.
https://bit.ly/3sw9q8U

Exertion-aware Path Generation - Crossminds We propose a novel approach for generating paths with desired exertion properties, which can be used for delivering highly realistic and immersive virtual reality applications that help users achieve exertion goals. Given a terrain as input, our optimization-based approach automatically generates fe...

[New Paper] Time-Travel Rephotography by researchers from UW Reality Lab University of Washington UC Berkeley and Google AI.
https://bit.ly/35YTqmr

Time-Travel Rephotography - Crossminds Many historical people are captured only in old, faded, black and white photos, that have been distorted by the limitations of early cameras and the passage of time. This paper simulates traveling back in time with a modern camera to rephotograph famous subjects. Unlike conventional image restoratio...

New Talk] HyperCon: Image-To-Video Model Transfer for Video-To-Video Translation Tasks by researchers from Computer Science and Engineering at Michigan and SamsungSemiUS.
https://bit.ly/3qgDS52

HyperCon: Image-To-Video Model Transfer for Video-To-Video Translation Tasks - Crossminds Video-to-video translation is more difficult than image-to-image translation due to the temporal consistency problem that, if unaddressed, leads to distracting flickering effects. Although video models designed from scratch produce temporally consistent results, training them to match the vast visua...

[New Paper] Learning Continuous Image Representation
with Local Implicit Image Function by researchers from UC San Diego and NVIDIA AI.

https://bit.ly/38DqcuK

Learning Continuous Image Representation with Local Implicit Image Function - Crossminds Project page: https://yinboc.github.io/liif/

[ Paper] FLAVR: A fast and efficient technique for video frame interpolation by researchers from UCSD CSE Department, Carnegie Mellon University School of Computer Science, and Facebook AI.
https://bit.ly/35uEfkl

FLAVR: A fast and efficient technique for video frame interpolation. - Crossminds This video summarizes our paper on "FLAVR: A Flow-Agnostic Video Representations for Fast Frame Interpolation". More details, source code, and paper can be found at https://tarun005.github.io/FLAVR/.

[ Blog] DALL·E: Creating Images from Text by researchers from OpenAI.
https://bit.ly/38raQK1

OpenAI DALL·E: Creating Images from Text (Blog Post Explained) - Crossminds OpenAI's newest model, DALL·E, shows absolutely amazing abilities in generating high-quality images from arbitrary text descriptions. Like GPT-3, the range of applications and the diversity of outputs is astonishing, given that this is a single model, trained on a purely auto...

[ Paper] Extracting Training Data from Large Language Models by researchers from Google AI Stanford Computer Science UC Berkeley Northeastern University Khoury College of Computer Sciences OpenAI Harvard John A. Paulson School of Engineering and Applied Sciences and Apple.
https://bit.ly/38lmV39

Extracting Training Data from Large Language Models (Paper Explained) - Crossminds This paper demonstrates a method to extract verbatim pieces of the training data from a trained language model. Moreover, some of the extracted pieces only appear a handful of times in the dataset. This points to serious security and privacy implications for models like GPT-3. The...

[ Paper] Automatic Validation of Textual Attribute Values in E-commerce Catalog by Learning with Limited Labeled Data by researchers from SUNY Buffalo State College and Amazon Science
https://bit.ly/3pX5KLc

[KDD2020] Automatic Validation of Textual Attribute Values in E-commerce Catalog by Learning with Limited Labeled Data - Crossminds Product catalogs are valuable resources for eCommerce website. In the catalog, a product is associated with multiple attributes whose values are short texts, such as product name, brand, functionality and flavor. Usually individual retailers self-report these key values, and thus the catalog informa...

[ Paper] Predictive Coding Approximates Backprop along Arbitrary Computation Graphs by researchers from Informatics, University of Edinburgh and University of Sussex.
https://bit.ly/3pR7VzT

Predictive Coding Approximates Backprop along Arbitrary Computation Graphs (Paper Explained) - Crossminds Backpropagation is the workhorse of modern deep learning and a core component of most frameworks, but it has long been known that it is not biologically plausible, driving a divide between neuroscience and machine learning. This paper shows that Predictive Coding, a much m...

[ Paper] Self-Distillation Amplifies Regularization in Hilbert Space by researchers from Google AI, DeepMind, UC Berkeley.
https://bit.ly/2L0Cu7D

Self-Distillation Amplifies Regularization in Hilbert Space - Crossminds Knowledge distillation introduced in the deep learning context is a method to transfer knowledge from one architecture to another. In particular, when the architectures are identical, this is called self-distillation. The idea is to feed in predictions of the trained model as new target values for r...

[ Talk] GAN-Leaks: A Taxonomy of Membership Inference Attacks against Generative Models by researchers from CISPA University of Maryland, College Park Max-Planck-Institut für Informatik (MPI-INF).
https://bit.ly/3pOVDIy

[CCS 2020] GAN-Leaks: A Taxonomy of Membership Inference Attacks against Generative Models - Crossminds GAN-Leaks: A Taxonomy of Membership Inference Attacks against Generative Models (CCS 2020) Dingfan Chen, Ning Yu, Yang Zhang, Mario Fritz Long Video More info: https://github.com/DingfanChen/GAN-Leaks

[Research Paper] A Hybrid Approach to Hypernym Discovery by researchers from Computer Research Institute of Montreal.
https://bit.ly/2X4HdHF

CRIM at SemEval-2018 Task 9: A Hybrid Approach to Hypernym Discovery | Research Paper Walkthrough - Crossminds A hypernym describes a more broader concept of the word, wherease, hyponym is more about defining finer granularity. In linguistics, a hyponym is a word or phrase whose semantic field is included within that of another word, its hyperonym or hypernym...

[ Talk] DronePose: Photorealistic UAV-Assistant Dataset Synthesis for 3D Pose Estimation by researchers from Visual Computing Lab, Information Technologies Institute, Centre for Research and Technology Hellas, Greece.
https://bit.ly/3548VJ4

[ECCV 2020] - DronePose: Photorealistic UAV-Assistant Dataset Synthesis for 3D Pose Estimation - Crossminds In this work we consider UAVs as cooperative agents supporting human users in their operations. In this context, the 3D localisation of the UAV assistant is an important task that can facilitate the exchange of spatial information between the user and the UAV. To address this in a data-driven manner...

[ Oral] Hierarchical Conditional Relation Networks for Video Question Answering by researchers from A²I² - Applied Artificial Intelligence Institute.
https://bit.ly/2Lb66Pb

Hierarchical Conditional Relation Networks for Video Question Answering - Oral Presentation - Crossminds Authors: Thao Minh Le, Vuong Le, Svetha Venkatesh, Truyen Tran Description: Video question answering (VideoQA) is challenging as it requires modeling capacity to distill dynamic visual artifacts and distant relations and to associate them with linguistic concepts. We introduce a general-purpose reus...

[ Paper] Node2Vec: Scalable Feature Learning for Networks by researchers from Stanford University.
https://bit.ly/3hxanZn

Node2Vec: Scalable Feature Learning for Networks | ML with Graphs | Research Paper Walkthrough - Crossminds In this video, we will walkthrough one of the foundational papers in the field of graph neural networks called Node2Vec, that tries to learn latent representation for nodes in the graph in a unsupervised way by using the notion of skip-gram algorithm applied...

[ Oral] Online Invariance Selection for Local Feature Descriptors by researchers from Informatik / Computer Science D-INFK, ETH Zürich and Microsoft Research.
https://bit.ly/2WTuyHN

[ECCV 2020 (Oral)] Online Invariance Selection for Local Feature Descriptors - Crossminds Paper presented at ECCV 2020 about an Online Invariance Selection for Local Feature Descriptors. We introduce a new method called LISRD able to leverage descriptors with different invariances and to perform an online selection of the most adapted invariance when matching two images. Paper: https://a...

[ Talk] Object Finding in Cluttered Scenes Using Interactive Perception by researchers from Gideon Brothers and Computer Vision Lab.
https://bit.ly/2Jsgl18

[ICRA 2020] Object Finding in Cluttered Scenes Using Interactive Perception - Crossminds Paper presented at ICRA 2020 about an Object Finding in Cluttered Scenes Using Interactive Perception. We propose a method to find hidden objects in a clutter of objects with interactive perception. A robot manipulator equipped with a camera can interact with the scene and leverages reinforcement le...

[ Talk] List-wise Fairness Criterion for Point Processes by researchers from LSU College of Science.
https://bit.ly/38GfeUo

[KDD 2020]List-wise Fairness Criterion for Point Processes - Crossminds Many types of event sequence data exhibit triggering and clustering properties in space and time. Point processes are widely used in modeling such event data with applications such as predictive policing and disaster event forecasting. Although current algorithms can achieve significant event predic...

[ Talk] Fast Video Object Segmentation using the Global Context Module by researchers from The University of Hong Kong - HKU - 香港大學.
https://bit.ly/3rDJdEP

ECCV 2020 - short presentation - Fast Video Object Segmentation using the Global Context Module - Crossminds 1-minute presentation on Fast Video Object Segmentation using the Global Context Module, an ECCV 2020 work by Tencent and The University of Hong Kong. Paper link: https://arxiv.org/abs/2001.11243​ TL;DR: We proposed a novel memory module for neural networks that effectively and efficiently propaga...

[ Talk] Active Search using Meta-Bandits by researchers from Language Technologies Institute at Carnegie Mellon University, UW-Madison Dept. of Computer Sciences, and Computer Science and Engineering at Lehigh University .
https://bit.ly/3pst4Af

CIKM2020 Active Search using Meta-Bandits - Crossminds Paper: https://dl.acm.org/doi/abs/10.1145/3340531.3417409

[ Talk] Efficient Attention: Attention with Linear Complexities by researchers from SenseTime Group and The Chinese University of Hong Kong 香港中文大學 - CUHK.
https://bit.ly/3hoeuai

WACV 2021 - Efficient Attention: Attention with Linear Complexities - Crossminds Presentation on Efficient Attention: Attention with Linear Complexities, a WACV 2021 work by SenseTime International and The Chinese University of Hong Kong. Paper link: https://arxiv.org/abs/1812.01243​ TL;DR: We proposed a novel attention mechanism with linear memory and computational complexiti...