Dmitry Kangin
YOU?
Author Swipe
View article: Transfer learning from inorganic materials to ivory detection
Transfer learning from inorganic materials to ivory detection Open
This paper describes the automatic identification of ivory using Raman spectroscopy data and deep neural network (DNN) models pre-trained on open-source data from inorganic minerals. The proposed approach uses transfer learning (TL) from f…
View article: i-WiViG: Interpretable Window Vision GNN
i-WiViG: Interpretable Window Vision GNN Open
Deep learning models based on graph neural networks have emerged as a popular approach for solving computer vision problems. They encode the image into a graph structure and can be beneficial for efficiently capturing the long-range depend…
View article: COMIX: Compositional Explanations using Prototypes
COMIX: Compositional Explanations using Prototypes Open
Aligning machine representations with human understanding is key to improving interpretability of machine learning (ML) models. When classifying a new image, humans often explain their decisions by decomposing the image into concepts and p…
View article: IMAFD: An Interpretable Multi-stage Approach to Flood Detection from time series Multispectral Data
IMAFD: An Interpretable Multi-stage Approach to Flood Detection from time series Multispectral Data Open
In this paper, we address two critical challenges in the domain of flood detection: the computational expense of large-scale time series change detection and the lack of interpretable decision-making processes on explainable AI (XAI). To o…
View article: Unsupervised Domain Adaptation within Deep Foundation Latent Spaces
Unsupervised Domain Adaptation within Deep Foundation Latent Spaces Open
The vision transformer-based foundation models, such as ViT or Dino-V2, are aimed at solving problems with little or no finetuning of features. Using a setting of prototypical networks, we analyse to what extent such foundation models can …
View article: On Neuron Activation Pattern and Applications
On Neuron Activation Pattern and Applications Open
As various deep learning applications have been deployed in diverse areas, the explainability of neural networks is becoming increasingly important in the research field. Besides being desirable on its own account, explainability also ofte…
View article: Towards interpretable-by-design deep learning algorithms
Towards interpretable-by-design deep learning algorithms Open
The proposed framework named IDEAL (Interpretable-by-design DEep learning ALgorithms) recasts the standard supervised classification problem into a function of similarity to a set of prototypes derived from the training data, while taking …
View article: Imbedding Deep Neural Networks
Imbedding Deep Neural Networks Open
Continuous-depth neural networks, such as Neural ODEs, have refashioned the understanding of residual neural networks in terms of non-linear vector-valued optimal control problems. The common solution is to use the adjoint sensitivity meth…
View article: Skillful Precipitation Nowcasting using Deep Generative Models of Radar
Skillful Precipitation Nowcasting using Deep Generative Models of Radar Open
Precipitation nowcasting, the high-resolution forecasting of precipitation up to two hours ahead, supports the real-world socio-economic needs of many sectors reliant on weather-dependent decision-making. State-of-the-art operational nowca…
View article: Temporarily retracted: Reconsidering the Roman workshop: using computer vision to analyse the making of ancient inscriptions
Temporarily retracted: Reconsidering the Roman workshop: using computer vision to analyse the making of ancient inscriptions Open
This paper is currently not available due to a delay in the granting of permission for publishing the images. The paper will be made visible as soon as this permission is granted. We apologize for the inconvenience.
View article: A review of radar-based nowcasting of precipitation and applicable machine learning techniques
A review of radar-based nowcasting of precipitation and applicable machine learning techniques Open
A 'nowcast' is a type of weather forecast which makes predictions in the very short term, typically less than two hours - a period in which traditional numerical weather prediction can be limited. This type of weather prediction has import…
View article: On-Policy Trust Region Policy Optimisation with Replay Buffers
On-Policy Trust Region Policy Optimisation with Replay Buffers Open
Building upon the recent success of deep reinforcement learning methods, we investigate the possibility of on-policy reinforcement learning improvement by reusing the data from several consecutive policies. On-policy methods bring many ben…
View article: Aggregated Sparse Attention for Steering Angle Prediction
Aggregated Sparse Attention for Steering Angle Prediction Open
In this paper, we apply the attention mechanism to autonomous driving for steering angle prediction. We propose the first model, applying the recently introduced sparse attention mechanism to visual domain, as well as the aggregated extens…
View article: Continuous Control With a Combination of Supervised and Reinforcement Learning
Continuous Control With a Combination of Supervised and Reinforcement Learning Open
This is the author accepted manuscript. The final version is available from the Institute of Electrical and Electronics Engineers via the DOI in this record.
View article: Data fusion for unsupervised video object detection, tracking and geo-positioning
Data fusion for unsupervised video object detection, tracking and geo-positioning Open
In this work we describe a system and propose a novel algorithm for moving object detection and tracking based on video feed. Apart of many well-known algorithms, it performs detection in unsupervised style, using velocity criteria for the…
View article: Evolving Classifier TEDAClass for Big Data
Evolving Classifier TEDAClass for Big Data Open
In the era of big data, huge amounts of data are generated and updated every day, and their processing and analysis is an important challenge today. In order to tackle this challenge, it is necessary to develop specific techniques which ca…