SFI Visual Intelligence
SFI Visual Intelligence
  • 117
  • 41 151
Anomaly Detection with Conditioned Denoising Diffusion Models: A. Mousakhan (University of Freiburg)
Arian Mousakhan, a PhD student at the Computer Vision Lab at the University of Freibrug, gave a presentation titled "Anomaly Detection with Conditioned Denoising Diffusion Models" on October 10th 2024 as part of the Visual Intelligence Online Seminar series.
Abstract:
Traditional reconstruction-based visual anomaly detection methods struggle to achieve competitive performance. This is primarily because most existing approaches are unable to precisely reconstruct anomalous inputs resulting in restored pattern that diverge from the original image. Moreover, methods often fail to conduct a robust comparison between the reconstructed and input images. In this paper, we propose Denoising Diffusion Anomaly Detection (DDAD) whereby a generic diffusion model is first trained only on nominal data. During inference, the reverse process is conditioned on the unperturbed input image by correcting the predicted noise at each denoising step. This novel mechanism accurately reconstructs anomalous regions while preserving the in-distribution patterns of the image. Our validate demonstrates the efficacy of DDAD on various datasets, achieving state-of-the-art results
Переглядів: 162

Відео

Benefits of Anatomical Motion Mode Imaging in LV Automatic Measurement: Durgesh K. Singh (UiT)
Переглядів 49Місяць тому
Durgesh Kumar Singh, a PhD student at UiT Machine Learning Group, gave a presentation titled "Benefits of Anatomical Motion Mode Imaging in LV Automatic Measurement" on August 29th 2024 as part of the Visual Intelligence Online Seminar series. Abstract: Linear measurements of the left ventricle (LV) are crucial for evaluating heart function and are typically performed using echocardiography to ...
FreqRISE: Explaining time series using frequency masking: Thea Brüsch (DTU Compute)
Переглядів 1273 місяці тому
Thea Brüsch, a PhD student at the Technical University of Denmark (DTU Compute) and guest researchers visiting Visual Intelligence at UiT, gave a presentation titled "FreqRISE: Explaining time series using frequency masking" on June 20th 2024 as part of the Visual Intelligence Online Seminar series. Abstract: Time series data is fundamentally important for describing many critical domains such ...
Annotation-Free Feature Learning for Improved Acoustic Target Classification: Ahmet Pala (UiB)
Переглядів 2864 місяці тому
Ahmet Pala, a PhD Research Fellow at the University of Bergen, gave a presentation titled "Towards Transforming Healthcare with General-Purpose AI on June 6th as part of the Visual Intelligence Online Seminar series. Abstract: Acoustic surveys are an important source of data for fisheries management. During the surveys, ship-mounted echo sounders send acoustic signals into the water and measure...
Towards Explainable AI 2.0 with Concept-based Explanations: Reduan Achtibat & Maxmilian Dreyer
Переглядів 2354 місяці тому
Reduan Achtibat and Maximilian Dreyer, two PhD students at Fraunhofer Heinrich Hertz Institute Berlin, gave a talk titled "Towards Explainable AI 2.0 with Concept-based Explanations" on May 28th 2024. The talk is divided into two parts: 0:00: Concept Relevance Propagation (CRP) 16:38: Prototypical Concept-based Explanations (PCX) Abstract: While local XAI methods explain individual predictions ...
Towards Transforming Healthcare with General-Purpose AI: Paul Jaeger (DKFZ/Helmholtz Imaging)
Переглядів 734 місяці тому
Paul Jaeger, Research Group Leader at the German Cancer Research Center (DKFZ) and Helmholtz Imaging, gave a presentation titled "Towards Transforming Healthcare with General-Purpose AI on May 23rd 2024 as part of the Visual Intelligence Online Seminar series. Abstract: In this talk, I will outline the current challenges of leveraging AI for clinical impact at scale. Focusing on medical image a...
Explainability Heatmaps for Unsupervised Group Robustness to Spurious Correlations: R. Chakraborty
Переглядів 1155 місяців тому
Rwiddhi Chakraborty, a doctoral research fellow at UiT Machine Learning Group, gave a presentation titled "ExMap: Leveraging Explainability Heatmaps for Unsupervised Group Robustness to Spurious Correlations" in our Visual Intelligence Online Seminar series (April 25th 2024). Abstract: Group robustness strategies aim to mitigate learned biases in deep learning models that arise from spurious co...
Earth Observation Foundation Models and their Applications: Casper Fibæk (The ESA Phi-Lab)
Переглядів 2086 місяців тому
Casper Fibæk, a Research Fellow at The ESA Phi-Lab, gave a presentation titled "Earth Observation Foundation Models and their Applications" (April 11th 2024) Abstract: At the ESA-Philab, we are developing and testing a range of foundation models, and particularly the effect of different pretraining tasks on downstream applications. Casper will present both the work on the models, tasks, as well...
Generative AI for High Quality 2D and 3D Echo Images: Cristiana Tiago (GE Healthcare)
Переглядів 1807 місяців тому
Cristiana Tiago, data scientist at GE Healthcare, gave a presentation titled "Generative AI for High Quality 2D and 3D Echo Images" (14th March 2024) Abstract: In cardiovascular imaging, the integration of Generative Artificial Intelligence holds great promise for enhancing the quality and availability of diagnostic tools. This presentation explores the application of Generative AI,specifically...
Achieving Data-Efficient Neural Networks with Hybrid Concept-based Models: Tobias Opsahl (UiO)
Переглядів 1598 місяців тому
Tobias Opsahl, a master's student in data science at Institute of Mathematics, University of Oslo, gave a presentation titled "Achieving Data-efficient Neural Networks with Hybrid Concept-based Models" (15th Feb. 2024) Abstract: Most datasets used for machine learning consist of a single label per data point, which is used to optimise the model. However, in cases where more information than jus...
XAI Generated Blind-masks for Self-Supervised Seismic Denoising: Claire Birnie (KAUST)
Переглядів 1128 місяців тому
Dr. Claire Birnie, research scientist at King Abdullah University of Science and Technology, gave a presentation titled "XAI Generated Blind-masks for Self-Supervised Seismic Denoising" (1 Feb. 2024) Abstract: Self-supervised denoising overcomes the challenge posed by traditional deep learning's requirement of clean training labels - an unobtainable requirement for field seismic data. I will be...
Accelerated Deep Learning via High-performance Computing: Razick, Agueny, Bjørve(Norwegian AI Cloud)
Переглядів 559 місяців тому
1. Welcome to HPC-intro tutorial (Sabry Razick UiO/NRIS) - The following link leads to the presentation training.pages.sigma2.no/tutorials/gpus-on-hpc/ 2. Accelerated Deep Learning on the supercomputer LUMI-G Part I: Hicham Agueny (UiB/NRIS) I.1-Introduction to supercomputer LUMI I.2-Basics of AMD-GPU topology I.3-Distributed Deep Learning with Horovod-TensorFlow on LUMI-G Part II: Magnar Bjørg...
Fair Self-supervised Learning in Multiple Modalities with Applications to Medicine: N. Razavian(NYU)
Переглядів 929 місяців тому
Narges Razavian, an associate professor at the New York University Langone Health, provided a keynote for the NLDL conference 2024 (11 Jan 2024). Title: Fair Self supervised Learning in multiple modalities (Imaging, EHR, and in combination) with Applications to Medicine Abstract: Recent progress in self-supervised learning (SSL), and availability of large clinical datasets that include millions...
How Do We Ensure That Trustworthy AI Remains Trustworthy? Aase Feragen (Technical Univ. of Denmark)
Переглядів 569 місяців тому
Aase Feragen, a professor at the Technical University of Denmark (DTU), provided a keynote for the NLDL conference 2024 (10 Jan 2024). Title: How do we ensure that Trustworthy AI remains trustworthy? Abstract: "Trustworthy AI" encompasses a range of approaches designed to promote safe and responsible use of AI. Prominent subfields include algorithmic fairness, explainable AI (XAI), and uncertai...
Reliability of Artificial Intelligence - Chances and Challenges: Gitta Kutyniok (LMU Munich)
Переглядів 1129 місяців тому
Gitta Kutyniok, a professor at Ludwig-Maximilian University in Munich, provided a keynote to the NLDL conference 2024 (10 Jan 2024). Title: Reliability of Artificial Intelligence: Chances and Challenges Abstract: Artificial intelligence is currently leading to one breakthrough after the other, both in public life with, for instance, autonomous driving and speech recognition, and in the sciences...
The Statistical Finite Element Method: Mark Girolami (Univ. of Cambridge/ The Alan Turing Institute)
Переглядів 1299 місяців тому
The Statistical Finite Element Method: Mark Girolami (Univ. of Cambridge/ The Alan Turing Institute)
Coding a Diffusion Model from Scratch: Filippo Maria Bianchi (UiT The Arctic University of Norway)
Переглядів 2679 місяців тому
Coding a Diffusion Model from Scratch: Filippo Maria Bianchi (UiT The Arctic University of Norway)
Learning from Limited Labeled Data for Few-shot Medical Image Segmentation: Michael Kampffmeyer(UiT)
Переглядів 3009 місяців тому
Learning from Limited Labeled Data for Few-shot Medical Image Segmentation: Michael Kampffmeyer(UiT)
Discriminative Multimodal Learning via Conditional Priors in Generative Models: Robert Jenssen (UiT)
Переглядів 569 місяців тому
Discriminative Multimodal Learning via Conditional Priors in Generative Models: Robert Jenssen (UiT)
Trustworthy and Fair AI: Srishti Gautam (UiT The Arctic University of Norway)
Переглядів 1819 місяців тому
Trustworthy and Fair AI: Srishti Gautam (UiT The Arctic University of Norway)
Assessing an AI System’s Compliance with the Laws of Decision-Making: Mathias K. Hauglid (SPKI/UNN)
Переглядів 7410 місяців тому
Assessing an AI System’s Compliance with the Laws of Decision-Making: Mathias K. Hauglid (SPKI/UNN)
Representation Learning for Multimodal Image Registration: Elisabeth Wetzer (Karolinska Inst. & UiT)
Переглядів 24811 місяців тому
Representation Learning for Multimodal Image Registration: Elisabeth Wetzer (Karolinska Inst. & UiT)
FullFormer - Generating Shapes Inside Shapes: Tejaswini Medi (University of Siegen)
Переглядів 14511 місяців тому
FullFormer - Generating Shapes Inside Shapes: Tejaswini Medi (University of Siegen)
Data-informed Distributions for Sampling Neural Network Weights: Erik Bolager (TU Munich)
Переглядів 90Рік тому
Data-informed Distributions for Sampling Neural Network Weights: Erik Bolager (TU Munich)
Recent Deep Learning Models and their Applicability in the Marine Domain: Changkyu Choi (UiT)
Переглядів 103Рік тому
Recent Deep Learning Models and their Applicability in the Marine Domain: Changkyu Choi (UiT)
Graph-Based Reasoning for Dialogue Models: Nicholas Walker (Norsk Regnesentral)
Переглядів 73Рік тому
Graph-Based Reasoning for Dialogue Models: Nicholas Walker (Norsk Regnesentral)
Zero-Waste Machine Learning: Tomasz Trzciński (Warsaw University of Technology)
Переглядів 188Рік тому
Zero-Waste Machine Learning: Tomasz Trzciński (Warsaw University of Technology)
NLDL 2023 Keynote: "Learning to Read Xray", Polina Golland (MIT CSAIL)
Переглядів 205Рік тому
NLDL 2023 Keynote: "Learning to Read Xray", Polina Golland (MIT CSAIL)
NLDL 2023 Keynote:"Deep Learning and Remote Sensing for Ecosystem Monitoring", Christian Igel (DIKU)
Переглядів 94Рік тому
NLDL 2023 Keynote:"Deep Learning and Remote Sensing for Ecosystem Monitoring", Christian Igel (DIKU)
NLDL 2023 Keynote: "AI for Science", Mihaela van der Schaar (University of Cambridge)
Переглядів 227Рік тому
NLDL 2023 Keynote: "AI for Science", Mihaela van der Schaar (University of Cambridge)

КОМЕНТАРІ

  • @ЮлианДьячков
    @ЮлианДьячков 6 днів тому

    Metz Street

  • @LawannaYosten-p8w
    @LawannaYosten-p8w 17 днів тому

    Maddison Plain

  • @DorothyTinker-t6c
    @DorothyTinker-t6c 17 днів тому

    Rae Place

  • @JulieHeidrick-h3z
    @JulieHeidrick-h3z 19 днів тому

    Ortiz Springs

  • @GeorgeSchultz-f5s
    @GeorgeSchultz-f5s 20 днів тому

    Eloisa Ridge

  • @ClintonEnglish-c7b
    @ClintonEnglish-c7b 23 дні тому

    Scot Brooks

  • @BradMunteanu-k5m
    @BradMunteanu-k5m 24 дні тому

    Medhurst Trail

  • @MaudKelly-u9i
    @MaudKelly-u9i 27 днів тому

    Giovanny Burgs

  • @WillettaShenkel-u8g
    @WillettaShenkel-u8g 27 днів тому

    Eugenia Inlet

  • @MariaBanks-t7p
    @MariaBanks-t7p 27 днів тому

    Josh Plains

  • @AnnaWymer-v5k
    @AnnaWymer-v5k 28 днів тому

    Eliseo Crossroad

  • @DorisYarbrough-p9p
    @DorisYarbrough-p9p Місяць тому

    Mayert Causeway

  • @EliseShaw-q4p
    @EliseShaw-q4p Місяць тому

    Oberbrunner Groves

  • @ScottBickham-z8r
    @ScottBickham-z8r Місяць тому

    Hyatt Islands

  • @SamuelSnider-l7l
    @SamuelSnider-l7l Місяць тому

    Will Expressway

  • @AnnaRodriguez-j1g
    @AnnaRodriguez-j1g Місяць тому

    Gage Stream

  • @BennieAyala-i7u
    @BennieAyala-i7u Місяць тому

    Bradtke Ports

  • @ferdinandcoleman7594
    @ferdinandcoleman7594 Місяць тому

    Rippin Squares

  • @DebraBaxter-w1c
    @DebraBaxter-w1c Місяць тому

    Christiansen Cove

  • @DanielKell-s8o
    @DanielKell-s8o Місяць тому

    Maya Falls

  • @AmeliaAlva-d9v
    @AmeliaAlva-d9v Місяць тому

    720 Von Union

  • @imadsaddik
    @imadsaddik 4 місяці тому

    Thanks for sharing

  • @DarrenR80
    @DarrenR80 4 місяці тому

    Hi Jon, your talk about deep learning algorithms was interesting this is something I would like to learn more about. Did you apply your experimental models to the k space domain by regressing the images back into the data domain or were you able to use the raw data. I am concern with the pixelated under sampling (discussed at time point~5 min and 40s) is this is not something we could reproduce clinically. We can only fill k space linearly, concentrically or radially. The parallel undersampling is something we could mimic as we can opt to reduce k space filling in the phase encoding direction. I look forward to your response Kind regards Darren

  • @takwabenlaribi
    @takwabenlaribi 5 місяців тому

    Please can you send me the code for model . Thanks

  • @N1sr0c
    @N1sr0c 5 місяців тому

  • @jonathanjonathansen
    @jonathanjonathansen 7 місяців тому

    Inspiring talk, made me see things in a different way!

  • @ghadialhajj
    @ghadialhajj 8 місяців тому

    Nice topic, and very well presented, Tobias!

  • @MDNQ-ud1ty
    @MDNQ-ud1ty 8 місяців тому

    How does any of this stop the psychopaths taking over humanity from destroying it? Seems to be helping them IMO.

  • @AkshayPimpalkar924
    @AkshayPimpalkar924 10 місяців тому

    Thank you for the talk. It was super interesting. Cheers!

  • @Vetrivel.Shanmugam
    @Vetrivel.Shanmugam 11 місяців тому

    How to build this shape shifter concept using neural network ? ua-cam.com/video/ddFVypwsFSw/v-deo.html

  • @Vetrivel.Shanmugam
    @Vetrivel.Shanmugam 11 місяців тому

    Can you make a video about probabilistic functions and artificial neural networks? why to specifically take insipiration from neural network to solve probabilisting problrms ?

  • @megankoss
    @megankoss Рік тому

    "PromoSM"

  • @RalphDratman
    @RalphDratman Рік тому

    I find it difficult to understand the speaker's voice.

    • @Qwertyuiopasdfghjklzxcvbnm266
      @Qwertyuiopasdfghjklzxcvbnm266 Рік тому

      Thanks for the comment. I did not realize this while speaking, but I agree with your comment. The location of the mic should be readjusted for the next presentation. Hope the generated subtitle helps.

    • @RalphDratman
      @RalphDratman Рік тому

      @@Qwertyuiopasdfghjklzxcvbnm266 Thank you for your reply. Yes, the subtitles help a lot.

  • @chongduan8887
    @chongduan8887 Рік тому

    Great work!

  • @sahhaf1234
    @sahhaf1234 Рік тому

    @26:55 what is Y-hat????

  • @sahhaf1234
    @sahhaf1234 Рік тому

    I agree that this is a very good talk, and subtitles will help it to reach a wider audience...

  • @lszuo
    @lszuo Рік тому

    really inspiring talk. hope the ppt could be uploaded as well. cheers!

  • @genericperson8238
    @genericperson8238 Рік тому

    I love this professor's presentation but it can sometimes be difficult to understand what he is saying. It would really help if there was a written transcript of the presentation.

  • @rikkinevers9616
    @rikkinevers9616 2 роки тому

    Pᵣₒmₒˢᵐ

  • @missword7912
    @missword7912 2 роки тому

    Look forward to more sessions about your current research, Dr. Yu. 🤩

  • @Laszer271
    @Laszer271 2 роки тому

    Very interesting talk. Are there any easy to understand sources so I can learn more about the approaches Raul mentioned or about the field in general? So far I have always ignored noisiness of my data when training models.

  • @Vikram-wx4hg
    @Vikram-wx4hg 2 роки тому

    Thanks for a very nice presentation on the topic!

  • @evamorisot2592
    @evamorisot2592 2 роки тому

    slay kween

  • @Laszer271
    @Laszer271 2 роки тому

    What sense does image segmentation make as a generative model problem? I imagine you could translate any problem into a generative model problem. E.g. instead of classification we can do generation of class conditioned on input image. Does formulating it in such a way makes more sense for image segmentation than for image classification or was that the point that any problem can be formulated this way?

  • @beniaminojewkes8027
    @beniaminojewkes8027 2 роки тому

    🅿🆁🅾🅼🅾🆂🅼

  • @levanhieu327
    @levanhieu327 2 роки тому

    This is what i need right now, thank you somuch

  • @vi5hnupradeep
    @vi5hnupradeep 2 роки тому

    Thank you so much

  • @surajaatreya4274
    @surajaatreya4274 2 роки тому

    Thanks for the nice talk!!

  • @chapmanezinga9652
    @chapmanezinga9652 2 роки тому

    Merci

  • @agneafaith1939
    @agneafaith1939 3 роки тому

    Am happy I finally got a permanent cure to my Herpes virus after taking the the medication I got from Dr oyakhire.