We study the problem of fitting a model to a dynamical environment when new modes of behavior emerge sequentially. The learning model is aware when a new mode appears, but it does not have access to the true modes of individual training sequences. We devise a novel continual learning method that maintains a descriptor of the mode of an encountered sequence in a neural episodic memory. We employ a Dirichlet Process prior on the attention weights of the memory to foster efficient storage of the mode descriptors. Our method performs continual learning by transferring knowledge across tasks by retrieving the descriptors of similar modes of past tasks to the mode of a current sequence and feeding this descriptor into its transition kernel as control input. We observe the continual learning performance of our method to compare favorably to the mainstream parameter transfer approach.
@inproceedings{akgul2022cddp,title={Continual Learning of Multi-modal Dynamics with External Memory},author={Akgül, A. and Unal, G. and Kandemir, M.},booktitle={Learning for Dynamics and Control Conference (L4DC)},year={2024},}
ICLR
Calibrating Bayesian UNet++ for Sub-Seasonal Forecasting
B. Asan, A. Akgül, A. Unal, and 2 more authors
In Workshop on Tackling Climate Change with Machine Learning 2024
Seasonal forecasting is a crucial task when it comes to detecting the extreme heat and colds that occur due to climate change. Confidence in the predictions should be reliable since a small increase in the temperatures in a year has a big impact on the world. Calibration of the neural networks provides a way to ensure our confidence in the predictions. However, calibrating regression models is an under-researched topic, especially in forecasters. We calibrate a UNet++ based architecture, which was shown to outperform physics-based models in temperature anomalies. We show that with a slight trade-off between prediction error and calibration error, it is possible to get more reliable and sharper forecasts. We believe that calibration should be an important part of safety-critical machine learning applications such as weather forecasters.
@inproceedings{asan2024calibrating,title={Calibrating Bayesian UNet++ for Sub-Seasonal Forecasting},author={Asan, B. and Akgül, A. and Unal, A. and Kandemir, M. and Unal, G.},booktitle={Workshop on Tackling Climate Change with Machine Learning},year={2024},}
arXiv
Probabilistic Actor-Critic: Learning to Explore with PAC-Bayes Uncertainty
We introduce Probabilistic Actor-Critic (PAC), a novel reinforcement learning algorithm with improved continuous control performance thanks to its ability to mitigate the exploration-exploitation trade-off. PAC achieves this by seamlessly integrating stochastic policies and critics, creating a dynamic synergy between the estimation of critic uncertainty and actor training. The key contribution of our PAC algorithm is that it explicitly models and infers epistemic uncertainty in the critic through Probably Approximately Correct-Bayesian (PAC-Bayes) analysis. This incorporation of critic uncertainty enables PAC to adapt its exploration strategy as it learns, guiding the actor’s decision-making process. PAC compares favorably against fixed or pre-scheduled exploration schemes of the prior art. The synergy between stochastic policies and critics, guided by PAC-Bayes analysis, represents a fundamental step towards a more adaptive and effective exploration strategy in deep reinforcement learning. We report empirical evaluations demonstrating PAC’s enhanced stability and improved performance over the state of the art in diverse continuous control problems.
@article{tasdighi2024probabilistic,title={Probabilistic Actor-Critic: Learning to Explore with PAC-Bayes Uncertainty},author={Tasdighi, B. and Werge, N. and Wu, Y. and Kandemir, M.},booktitle={arXiv Preprint},year={2024},}
2023
arXiv
EdVAE: Mitigating Codebook Collapse with Evidential Discrete Variational Autoencoders
Codebook collapse is a common problem in training deep generative models with discrete representation spaces like Vector Quantized Variational Autoencoders (VQ-VAEs). We observe that the same problem arises for the alternatively designed discrete variational autoencoders (dVAEs) whose encoder directly learns a distribution over the codebook embeddings to represent the data. We hypothesize that using the softmax function to obtain a probability distribution causes the codebook collapse by assigning overconfident probabilities to the best matching codebook elements. In this paper, we propose a novel way to incorporate evidential deep learning (EDL) instead of softmax to combat the codebook collapse problem of dVAE. We evidentially monitor the significance of attaining the probability distribution over the codebook embeddings, in contrast to softmax usage. Our experiments using various datasets show that our model, called EdVAE, mitigates codebook collapse while improving the reconstruction performance, and enhances the codebook usage compared to dVAE and VQ-VAE based models.
@inproceedings{baykal2023edvae,author={Baykal, G. and Kandemir, M. and Unal, G.},booktitle={arXiv Preprint},year={2023},}
NeurIPS
Improved Algorithms for Stochastic Linear Bandits Using Tail Bounds for Martingale Mixtures
We present improved algorithms with worst-case regret guarantees for the stochastic linear bandit problem. The widely used "optimism in the face of uncertainty" principle reduces a stochastic bandit problem to the construction of a confidence sequence for the unknown reward function. The performance of the resulting bandit algorithm depends on the size of the confidence sequence, with smaller confidence sets yielding better empirical performance and stronger regret guarantees. In this work, we use a novel tail bound for adaptive martingale mixtures to construct confidence sequences which are suitable for stochastic bandits. These confidence sequences allow for efficient action selection via convex programming. We prove that a linear bandit algorithm based on our confidence sequences is guaranteed to achieve competitive worst-case regret. We show that our confidence sequences are tighter than competitors, both empirically and theoretically. Finally, we demonstrate that our tighter confidence sequences give improved performance in several hyperparameter tuning tasks.
@inproceedings{flynn2023improved,author={Flynn, H. and Reeb, D. and Kandemir, M. and Peters, J.},year={2023},title={Improved Algorithms for Stochastic Linear Bandits Using Tail Bounds for Martingale Mixtures},booktitle={Neural Information Processing Systems},}
ACML
Estimation of Counterfactual Interventions under Uncertainties
J. Weilbach, S. Gerwinn, M. Kandemir, and 1 more author
Counterfactual analysis is intuitively performed by humans on a daily basis eg. "What should I have done differently to get the loan approved?". Such counterfactual questions also steer the formulation of scientific hypotheses. More formally it provides insights about potential improvements of a system by inferring the effects of hypothetical interventions into a past observation of the system’s behaviour which plays a prominent role in a variety of industrial applications. Due to the hypothetical nature of such analysis, counterfactual distributions are inherently ambiguous. This ambiguity is particularly challenging in continuous settings in which a continuum of explanations exist for the same observation. In this paper, we address this problem by following a hierarchical Bayesian approach which explicitly models such uncertainty. In particular, we derive counterfactual distributions for a Bayesian Warped Gaussian Process thereby allowing for non-Gaussian distributions and non-additive noise. We illustrate the properties our approach on a synthetic and on a semi-synthetic example and show its performance when used within an algorithmic recourse downstream task.
@inproceedings{weilbach2023estimation,title={Estimation of Counterfactual Interventions under Uncertainties},author={Weilbach, J. and Gerwinn, S. and Kandemir, M. and Fraenzle, M.},booktitle={Asian Conference on Machine Learning},year={2023},}
T-PAMI
PAC-Bayes Bounds for Bandit Problems: A Survey and Experimental Comparison
H. Flynn, D. Reeb, M. Kandemir, and 1 more author
IEEE Transactions on Pattern Analysis and Machine Intelligence 2023
PAC-Bayes has recently re-emerged as an effective theory with which one can derive principled learning algorithms with tight performance guarantees. However, applications of PAC-Bayes to bandit problems are relatively rare, which is a great misfortune. Many decision-making problems in healthcare, finance and natural sciences can be modelled as bandit problems. In many of these applications, principled algorithms with strong performance guarantees would be very much appreciated. This survey provides an overview of PAC-Bayes performance bounds for bandit problems and an experimental comparison of these bounds. Our experimental comparison has revealed that available PAC-Bayes upper bounds on the cumulative regret are loose, whereas available PAC-Bayes lower bounds on the expected reward can be surprisingly tight. We found that an offline contextual bandit algorithm that learns a policy by optimising a PAC-Bayes bound was able to learn randomised neural network polices with competitive expected reward and non-vacuous performance guarantees.
@article{flyn2022pacbayes,title={PAC-Bayes Bounds for Bandit Problems: A Survey and Experimental Comparison},author={Flynn, H. and Reeb, D. and Kandemir, M. and Peters, J.},journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},year={2023},}
arXiv
BOF-UCB: A Bayesian-Optimistic Frequentist Algorithm for Non-Stationary Contextual Bandits
We propose a novel Bayesian-Optimistic Frequentist Upper Confidence Bound (BOF-UCB) algorithm for stochastic contextual linear bandits in non-stationary environments. This unique combination of Bayesian and frequentist principles enhances adaptability and performance in dynamic settings. The BOF-UCB algorithm utilizes sequential Bayesian updates to infer the posterior distribution of the unknown regression parameter, and subsequently employs a frequentist approach to compute the Upper Confidence Bound (UCB) by maximizing the expected reward over the posterior distribution. We provide theoretical guarantees of BOF-UCB’s performance and demonstrate its effectiveness in balancing exploration and exploitation on synthetic datasets and classical control tasks in a reinforcement learning setting. Our results show that BOF-UCB outperforms existing methods, making it a promising solution for sequential decision-making in non-stationary environments.
@article{werge2023bof,title={BOF-UCB: A Bayesian-Optimistic Frequentist Algorithm for Non-Stationary Contextual Bandits},author={Werge, N. and Akgül, A. and Kandemir, M.},journal={arXiv Preprint},year={2023},}
MDPI
ALReg: Registration of 3D Point Clouds Using Active Learning
Y.H. Sahin, O. Karabacak, M. Kandemir, and 1 more author
After the success of deep learning in point cloud segmentation and classification tasks, it has also been adopted as common practice in point cloud registration applications. State-of-the-art point cloud registration methods generally deal with this problem as a regression task to find the underlying rotation and translation between two point clouds. However, given two point clouds, the transformation between them could be calculated using only definitive point subsets from each cloud. Furthermore, training time is still a major problem among the current registration networks, whereas using a selective approach to define the informative point subsets can lead to reduced network training times. To that end, we developed ALReg, an active learning procedure to select a limited subset of point clouds to train the network. Each of the point clouds in the training set is divided into superpoints (small pieces of each cloud) and the training process is started with a small amount of them. By actively selecting new superpoints and including them in the training process, only a prescribed amount of data is used, hence the time needed to converge drastically decreases. We used DeepBBS, FMR, and DCP methods as our baselines to prove our proposed ALReg method. We trained DeepBBS and DCP on the ModelNet40 dataset and FMR on the 7Scenes dataset. Using 25% of the training data for ModelNet and 4% for the 7Scenes, better or similar accuracy scores are obtained in less than 20% of their original training times. The trained models are also tested on the 3DMatch dataset and better results are obtained than the original FMR training procedure.
@article{sahin2023alreg,title={ALReg: Registration of 3D Point Clouds Using Active Learning},author={Sahin, Y.H. and Karabacak, O. and Kandemir, M. and Unal, G.},journal={MDPI Applied Sciences},year={2023},url={https://www.mdpi.com/2076-3417/13/13/7422},}
arXiv
PAC-Bayesian Soft Actor-Critic Learning
B. Tasdighi, A. Akgül, K.K. Brink, and 1 more author
Actor-critic algorithms address the dual goals of reinforcement learning, policy evaluation and improvement, via two separate function approximators. The practicality of this approach comes at the expense of training instability, caused mainly by the destructive effect of the approximation errors of the critic on the actor. We tackle this bottleneck by employing an existing Probably Approximately Correct (PAC) Bayesian bound for the first time as the critic training objective of the Soft Actor-Critic (SAC) algorithm. We further demonstrate that the online learning performance improves significantly when a stochastic actor explores multiple futures by critic-guided random search. We observe our resulting algorithm to compare favorably to the state of the art on multiple classical control and locomotion tasks in both sample efficiency and asymptotic performance.
@inproceedings{bahareh2022pac4sac,title={PAC-Bayesian Soft Actor-Critic Learning},author={Tasdighi, B. and Akgül, A. and Brink, K.K. and Kandemir, M.},booktitle={arXiv Preprint},year={2023},}
TMLR
Cheap and Deterministic Inference for Deep State-Space Models of Interacting Dynamical Systems
A. Look, B. Rakitsch, M. Kandemir, and 1 more author
Graph neural networks are often used to model interacting dynamical systems since they gracefully scale to systems with a varying and high number of agents. While there has been much progress made for deterministic interacting systems, modeling is much more challenging for stochastic systems in which one is interested in obtaining a predictive distribution over future trajectories. Existing methods are either computationally slow since they rely on Monte Carlo sampling or make simplifying assumptions such that the predictive distribution is unimodal. In this work, we present a deep state-space model which employs graph neural networks in order to model the underlying interacting dynamical system. The predictive distribution is multimodal and has the form of a Gaussian mixture model, where the moments of the Gaussian components can be computed via deterministic moment matching rules. Our moment matching scheme can be exploited for sample-free inference leading to more efficient and stable training compared to Monte Carlo alternatives. Furthermore, we propose structured approximations to the covariance matrices of the Gaussian components in order to scale up to systems with many agents. We benchmark our novel framework on two challenging autonomous driving datasets. Both confirm the benefits of our method compared to state-of-the-art methods. We further demonstrate the usefulness of our individual contributions in a carefully designed ablation study and provide a detailed empirical runtime analysis of our proposed covariance approximations.
@article{look2023cheap,title={Cheap and Deterministic Inference for Deep State-Space Models of Interacting Dynamical Systems},author={Look, A. and Rakitsch, B. and Kandemir, M. and Peters, J.},journal={Transactions on Machine Learning Research},year={2023},url={https://openreview.net/forum?id=dqgdBy4Uv5¬eId=xKtcWgwdxX},}
2022
NeurIPS
Learning Interacting Dynamical Systems with Latent Gaussian Process ODEs
We study for the first time uncertainty-aware modeling of continuous-time dynamics of interacting objects. We introduce a new model that decomposes independent dynamics of single objects accurately from their interactions. By employing latent Gaussian process ordinary differential equations, our model infers both independent dynamics and their interactions with reliable uncertainty estimates. In our formulation, each object is represented as a graph node and interactions are modeled by accumulating the messages coming from neighboring objects. We show that efficient inference of such a complex network of variables is possible with modern variational sparse Gaussian process inference techniques. We empirically demonstrate that our model improves the reliability of long-term predictions over neural network based alternatives and it successfully handles missing dynamic or static information. Furthermore, we observe that only our model can successfully encapsulate independent dynamics and interaction information in distinct functions and show the benefit from this disentanglement in extrapolation scenarios.
@inproceedings{yildiz2022learning,title={Learning Interacting Dynamical Systems with Latent Gaussian Process ODEs},author={Yildiz, C. and Kandemir, M. and Rakitsch, B.},booktitle={Neural Information Processing Systems},year={2022},}
ICLR
Evidential Turing Processes
M. Kandemir, A. Akgül, M. Haussmann, and 1 more author
In International Conference on Learning Representations 2022
A probabilistic classifier with reliable predictive uncertainties i) fits successfully to the target domain data, ii) provides calibrated class probabilities in difficult regions of the target domain (e.g. class overlap), and iii) accurately identifies queries coming out of the target domain and reject them. We introduce an original combination of Evidential Deep Learning, Neural Processes, and Neural Turing Machines capable of providing all three essential properties mentioned above for total uncertainty quantification. We observe our method on three image classification benchmarks to consistently improve the in-domain uncertainty quantification, out-of-domain detection, and robustness against input perturbations with one single model. Our unified solution delivers an implementation-friendly and computationally efficient recipe for safety clearance and provides intellectual economy to an investigation of algorithmic roots of epistemic awareness in deep neural nets.
@inproceedings{kandemir2022evidential,title={Evidential Turing Processes},author={Kandemir, M. and Akgül, A. and Haussmann, M. and Unal, G.},booktitle={International Conference on Learning Representations},year={2022},url={https://openreview.net/forum?id=84NMXTHYe-},}
T-PAMI
A Deterministic Approximation to Neural SDEs
A. Look, M. Kandemir, B. Rakitsch, and 1 more author
IEEE Transactions on Pattern Analysis and Machine Intelligence 2022
Neural Stochastic Differential Equations (NSDEs) model the drift and diffusion functions of a stochastic process as neural networks. While NSDEs are known to make accurate predictions, their uncertainty quantification properties haven been remained unexplored so far. We report the empirical finding that obtaining well-calibrated uncertainty estimations from NSDEs is computationally prohibitive. As a remedy, we develop a computationally affordable deterministic scheme which accurately approximates the transition kernel, when dynamics is governed by a NSDE. Our method introduces a bidimensional moment matching algorithm: vertical along the neural net layers and horizontal along the time direction, which benefits from an original combination of effective approximations. Our deterministic approximation of the transition kernel is applicable to both training and prediction. We observe in multiple experiments that the uncertainty calibration quality of our method can be matched by Monte Carlo sampling only after introducing high computation cost. Thanks to the numerical stability of deterministic training, our method also provides improvement in prediction accuracy.
@article{look2022adeterministic,title={A Deterministic Approximation to Neural SDEs},author={Look, A. and Kandemir, M. and Rakitsch, B. and Peters, J.},journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},year={2022},publisher={IEEE},}
L4DC
Traversing Time with Multi-Resolution Gaussian Process State-Space Models
K. Longi, J. Lindinger, O. Duennbier, and 3 more authors
In Learning for Dynamics and Control Conference 2022
Gaussian Process state-space models capture complex temporal dependencies in a principled manner by placing a Gaussian Process prior on the transition function. These models have a natural interpretation as discretized stochastic differential equations, but inference for long sequences with fast and slow transitions is difficult. Fast transitions need tight discretizations whereas slow transitions require backpropagating the gradients over long subtrajectories. We propose a novel Gaussian process state-space architecture composed of multiple components, each trained on a different resolution, to model effects on different timescales. The combined model allows traversing time on adaptive scales, providing efficient inference for arbitrarily long sequences with complex dynamics. We benchmark our novel method on semi-synthetic data and on an engine modeling task. In both experiments, our approach compares favorably against its state-of-the-art alternatives that operate on a single time-scale only.
@inproceedings{lungi2022gpssm,title={Traversing Time with Multi-Resolution Gaussian Process State-Space Models},author={Longi, K. and Lindinger, J. and Duennbier, O. and Kandemir, M. and Klami, A. and Rakitsch, B},booktitle={Learning for Dynamics and Control Conference},year={2022},}
DMKD
PAC-Bayesian lifelong learning for multi-armed bandits
We present a PAC-Bayesian analysis of lifelong learning. In the lifelong learning problem, a sequence of learning tasks is observed one-at-a-time, and the goal is to transfer information acquired from previous tasks to new learning tasks. We consider the case when each learning task is a multi-armed bandit problem. We derive lower bounds on the expected average reward that would be obtained if a given multi-armed bandit algorithm was run in a new task with a particular prior and for a set number of steps. We propose lifelong learning algorithms that use our new bounds as learning objectives. Our proposed algorithms are evaluated in several lifelong multi-armed bandit problems and are found to perform better than a baseline method that does not use generalisation bounds.
@article{flynn2022pac,title={PAC-Bayesian lifelong learning for multi-armed bandits},author={Flynn, H. and Reeb, D. and Kandemir, M. and Peters, J.},journal={Data Mining and Knowledge Discovery},year={2022},publisher={Springer},url={https://link.springer.com/article/10.1007/s10618-022-00825-4},}
2021
ICML
Inferring the Structure of Ordinary Differential Equations
J. Weilbach, S. Gerwinn, C. Weilbach, and 1 more author
@inproceedings{weilbach2021inferring,title={Inferring the Structure of Ordinary Differential Equations},author={Weilbach, J. and Gerwinn, S. and Weilbach, C. and Kandemir, M.},booktitle={ICML Time Series Workshop},year={2021},}
AISTATS
Learning Partially Known Stochastic Dynamics with Empirical PAC Bayes
M. Haussmann, S. Gerwinn, A. Look, and 2 more authors
In International Conference on Artificial Intelligence and Statistics 2021
@inproceedings{haussmann2021learning,title={Learning Partially Known Stochastic Dynamics with Empirical PAC Bayes},author={Haussmann, M. and Gerwinn, S. and Look, A. and Rakitsch, B. and Kandemir, M.},booktitle={International Conference on Artificial Intelligence and Statistics},year={2021},}
2020
NeurIPS
Differentiable Implicit Layers
A. Look, S. Doneva, M. Kandemir, and 2 more authors
@inproceedings{look2020differentiable,title={Differentiable Implicit Layers},author={Look, A. and Doneva, S. and Kandemir, M. and Gemulla, R. and Peters, J.},booktitle={NeurIPS ML for Engineering Workshop},year={2020},}
AABI
Bayesian Evidential Deep Learning with PAC Regularization
M. Haussmann, S. Gerwinn, and M. Kandemir
In Advances in Approximate Bayesian Inference Symposium 2020
@inproceedings{haussmann2020bayesian,title={Bayesian Evidential Deep Learning with PAC Regularization},author={Haussmann, M. and Gerwinn, S. and Kandemir, M.},booktitle={Advances in Approximate Bayesian Inference Symposium},year={2020},}
2019
IJCAI
Deep Active Learning with Adaptive Acquisition
M. Haußmann, F.A. Hamprecht, and M. Kandemir
In International Joint Conference on Artificial Intelligence 2019
@inproceedings{haussmann2019deep,title={Deep Active Learning with Adaptive Acquisition},author={Hau{\ss}mann, M. and Hamprecht, F.A. and Kandemir, M.},booktitle={International Joint Conference on Artificial Intelligence},year={2019},}
@inproceedings{look2019differential,title={Differential Bayesian Neural Nets},author={Look, A. and Kandemir, M.},booktitle={NeurIPS Bayesian Deep Learning Workshop},year={2019},}
UAI
Sampling-Free Variational Inference of Bayesian Neural Nets with Variance Backpropagation
M. Haussmann, F.A. Hamprecht, and M. Kandemir
In International Conference on Uncertainty in Artificial Intelligence 2019
@inproceedings{haussmann2019sampling,title={{Sampling-Free Variational Inference of Bayesian Neural Nets with Variance Backpropagation}},author={Haussmann, M. and Hamprecht, F.A. and Kandemir, M.},booktitle={International Conference on Uncertainty in Artificial Intelligence},pages={arXiv--preprint},year={2019},}
@article{kandemir2018variational,title={{Variational Closed-Form Deep Neural Net Inference}},author={Kandemir, M.},journal={Pattern Recognition Letters},volume={112},pages={145--151},year={2018},publisher={North-Holland},}
NeurIPS
Evidential Deep Learning to Quantify Classification Uncertainty
@inproceedings{sensoy2018evidential,title={Evidential Deep Learning to Quantify Classification Uncertainty},author={Sensoy, M. and Kaplan, L. and Kandemir, M.},booktitle={Neural Information Processing Systems},pages={3179--3189},year={2018},}
icccn
On Context-Aware DDoS Attacks using Deep Generative Networks
G. Gürsun, M. Sensoy, and M. Kandemir
In International Conference on Computer Communication and Networks (ICCCN) 2018
@inproceedings{gursun2018context,title={{On Context-Aware DDoS Attacks using Deep Generative Networks}},author={G{\"u}rsun, G. and Sensoy, M. and Kandemir, M.},booktitle={International Conference on Computer Communication and Networks (ICCCN)},pages={1--9},year={2018},organization={IEEE},}
@article{kandemir2018supervising,title={{Supervising Topic Models with Gaussian Processes}},author={Kandemir, M. and Keke{\c{c}}, T. and Yeniterzi, R.},journal={Pattern Recognition},volume={77},pages={226--236},year={2018},}
2017
CVPR
Variational Bayesian Multiple Instance Learning with Gaussian Processes
@inproceedings{haussmann2017variational,title={Variational Bayesian Multiple Instance Learning with Gaussian Processes},author={Hau{\ss}mann, M. and Hamprecht, F.A. and Kandemir, M.},booktitle={Computer Vision and Pattern Recognition},year={2017},}
NoF
Prediction of Active UE Number with Bayesian Neural Networks for Self-Organizing LTE Networks
O. Narmanlioglu, E. Zeydan, M. Kandemir, and 1 more author
In International Conference on the Network of the Future (NOF) 2017
@inproceedings{narmanlioglu2017prediction,title={Prediction of Active UE Number with Bayesian Neural Networks for Self-Organizing LTE Networks},author={Narmanlioglu, O. and Zeydan, E. and Kandemir, M. and Kranda, T.},booktitle={International Conference on the Network of the Future (NOF)},pages={73--78},year={2017},organization={IEEE},}
2016
ECCV
Gaussian Process Density Counting from Weak Supervision
M.v. Borstel, M. Kandemir, P. Schmidt, and 3 more authors
@inproceedings{borstel2016gaussian,title={Gaussian Process Density Counting from Weak Supervision},author={Borstel, M.v. and Kandemir, M. and Schmidt, P. and Rao, M.K. and Rajamani, K. and Hamprecht, F.A.},booktitle={European Conference on Computer Vision},pages={365--380},year={2016},organization={Springer, Cham},}
BMVC
Variational Weakly Supervised Gaussian Processes
M. Kandemir, M. Haußmann, F. Diego, and 3 more authors
@inproceedings{kandemir2016variational,title={Variational Weakly Supervised Gaussian Processes},author={Kandemir, M. and Hau{\ss}mann, M. and Diego, F. and Rajamani, K. and Van Der Laak, J. and Hamprecht, F.A.},booktitle={British Machine Vision Conference},pages={71--1},year={2016},}
Journal
Multiple Instance Learning: Robust Validation on Retinopathy of Prematurity
P. Rani, R. Elagiri, K. Rajamani, and 2 more authors
International Journal of Control Theory and Applicatioms 2016
@article{rani2016multiple,title={Multiple Instance Learning: Robust Validation on Retinopathy of Prematurity},author={Rani, P. and Elagiri, R. and Rajamani, K. and Kandemir, M. and Singh, D.},journal={International Journal of Control Theory and Applicatioms},volume={9},pages={451--459},year={2016},}
2015
journal
Computer-Aided Diagnosis from Weak Supervision: A Benchmarking Study
@article{kandemir2015computer,title={Computer-Aided Diagnosis from Weak Supervision: A Benchmarking Study},author={Kandemir, M. and Hamprecht, F.A.},journal={Computerized medical imaging and graphics},volume={42},pages={44--50},year={2015},}
miccai
Cell Event Detection in Phase-Contrast Microscopy Sequences from Few Annotations
M. Kandemir, C. Wojek, and F.A. Hamprecht
In International Conference on Medical Image Computing and Computer Assisted Interventions 2015
@inproceedings{kandemir2015cell,title={Cell Event Detection in Phase-Contrast Microscopy Sequences from Few Annotations},author={Kandemir, M. and Wojek, C. and Hamprecht, F.A.},booktitle={International Conference on Medical Image Computing and Computer Assisted Interventions},pages={316--323},year={2015},organization={Springer},}
ICACCI
Detection of Retinopathy of Prematurity using Multiple Instance Learning
P. Rani, E. Ramalingam Rajkumar, K. Rajamani, and 2 more authors
In International Conference on Advances in Computing, Communications and Informatics (ICACCI) 2015
@inproceedings{rani2015detection,title={Detection of Retinopathy of Prematurity using Multiple Instance Learning},author={Rani, P. and Rajkumar, E. Ramalingam and Rajamani, K. and Kandemir, M. and Singh, D.},booktitle={International Conference on Advances in Computing, Communications and Informatics (ICACCI)},pages={2233--2237},year={2015},organization={IEEE},}
Journal
Towards Brain-Activity-Controlled Information Retrieval: Decoding Image Relevance from MEG Signals
J.P. Kauppi, M. Kandemir, V.M. Saarinen, and 5 more authors
@article{kauppi2015towards,title={Towards Brain-Activity-Controlled Information Retrieval: Decoding Image Relevance from MEG Signals},author={Kauppi, J.P. and Kandemir, M. and Saarinen, V.M. and Hirvenkari, L. and Parkkonen, L. and Klami, A. and Hari, R. and Kaski, S.},journal={NeuroImage},volume={112},pages={288--298},year={2015},publisher={Elsevier},}
ICML
Asymmetric Transfer Learning with Deep Gaussian Processes
M. Kandemir
In International Conference on Machine Learning 2015
@inproceedings{kandemir2015asymmetric,title={Asymmetric Transfer Learning with Deep Gaussian Processes},author={Kandemir, M.},booktitle={International Conference on Machine Learning},pages={730--738},year={2015},organization={PMLR},}
PMLR
The Deep Feed-Forward Gaussian Process: An Effective Generalization to Covariance Priors
M. Kandemir, and F. A. Hamprecht
In Feature Extraction: Modern Questions and Challenges 2015
@inproceedings{kandemir2015deep,title={The Deep Feed-Forward Gaussian Process: An Effective Generalization to Covariance Priors},author={Kandemir, M. and Hamprecht, F. A.},booktitle={Feature Extraction: Modern Questions and Challenges},year={2015},}
2014
MICCAI
Empowering Multiple Instance Histopathology Cancer Diagnosis by Cell Graphs
M. Kandemir, C. Zhang, and F.A. Hamprecht
In International Conference on Medical Image Computing and Computer Assisted Interventions 2014
@inproceedings{kandemir2014empowering,title={Empowering Multiple Instance Histopathology Cancer Diagnosis by Cell Graphs},author={Kandemir, M. and Zhang, C. and Hamprecht, F.A.},booktitle={International Conference on Medical Image Computing and Computer Assisted Interventions},pages={228--235},year={2014},organization={Springer},}
ISBI
Digital Pathology: Multiple Instance Learning can Detect Barrett’s Cancer
M. Kandemir, A. Feuchtinger, A. Walch, and 1 more author
In International Symposium on Biomedical Imaging (ISBI) 2014
@inproceedings{kandemir2014digital,title={Digital Pathology: Multiple Instance Learning can Detect Barrett's Cancer},author={Kandemir, M. and Feuchtinger, A. and Walch, A. and Hamprecht, F.A.},booktitle={International Symposium on Biomedical Imaging (ISBI)},pages={1348--1351},year={2014},organization={IEEE},}
UAI
Instance Label Prediction by Dirichlet Process Multiple Instance Learning
M. Kandemir, and F. A. Hamprecht
In International Conference on Uncertainty in Artificial Intelligence 2014
@inproceedings{kandemir2014instance,title={Instance Label Prediction by Dirichlet Process Multiple Instance Learning},author={Kandemir, M. and Hamprecht, F. A.},booktitle={International Conference on Uncertainty in Artificial Intelligence},year={2014},}
MICCAI
Event Detection by Feature Unpredictability in Phase-Contrast Videos of Cell Cultures
M. Kandemir, J.C. Rubio, U. Schmidt, and 4 more authors
In International Conference on Medical Image Computing and Computer Assisted Interventions 2014
@inproceedings{kandemir2014event,title={Event Detection by Feature Unpredictability in Phase-Contrast Videos of Cell Cultures},author={Kandemir, M. and Rubio, J.C. and Schmidt, U. and Wojek, C. and Welbl, J. and Ommer, B. and Hamprecht, F.A.},booktitle={International Conference on Medical Image Computing and Computer Assisted Interventions},pages={154--161},year={2014},organization={Springer},}
Multiple Instance Learning with Response-Optimized Random Forests
C. Straehle, M. Kandemir, U. Koethe, and 1 more author
In International Conference on Pattern Recognition 2014
@inproceedings{straehle2014multiple,title={Multiple Instance Learning with Response-Optimized Random Forests},author={Straehle, C. and Kandemir, M. and Koethe, U. and Hamprecht, F.A.},booktitle={International Conference on Pattern Recognition},pages={3768--3773},year={2014},organization={IEEE},}
Journal
Multi-Task and Multi-View Learning of User State
M. Kandemir, A. Vetek, M. Gönen, and 2 more authors
@article{kandemir2014multi,title={Multi-Task and Multi-View Learning of User State},author={Kandemir, M. and Vetek, A. and G{\"o}nen, M. and Klami, A. and Kaski, S.},journal={Neurocomputing},volume={139},pages={97--106},year={2014},publisher={Elsevier},}
@inproceedings{kandemir2012learning,title={Learning Relevance from Natural Eye Movements in Pervasive Interfaces},author={Kandemir, M. and Kaski, S.},booktitle={International Conference on Multimodal Interaction},year={2012},}
ECML
Unsupervised Inference of Auditory Attention from Biosensors
M. Kandemir, A. Klami, A. Vetek, and 1 more author
In Joint European Conference on Machine Learning and Knowledge Discovery in Databases 2012
@inproceedings{kandemir2012unsupervised,title={Unsupervised Inference of Auditory Attention from Biosensors},author={Kandemir, M. and Klami, A. and Vetek, A. and Kaski, S.},booktitle={Joint European Conference on Machine Learning and Knowledge Discovery in Databases},pages={403--418},year={2012},organization={Springer},}
2011
ICONIP
Multitask learning using regularized multiple kernel learning
M. Gönen, M. Kandemir, and S. Kaski
In International Conference on Neural Information Processing 2011
@inproceedings{gonen2011multitask,title={Multitask learning using regularized multiple kernel learning},author={G{\"o}nen, M. and Kandemir, M. and Kaski, S.},booktitle={International Conference on Neural Information Processing},pages={500--509},year={2011},organization={Springer},}
journal
An Augmented Reality Interface to Contextual Information
A. Ajanki, M. Billinghurst, H. Gamper, and 8 more authors
@article{ajanki2011augmented,title={An Augmented Reality Interface to Contextual Information},author={Ajanki, A. and Billinghurst, M. and Gamper, H. and J{\"a}rvenp{\"a}{\"a}, T. and Kandemir, M. and Kaski, S. and Koskela, M. and Kurimo, M. and Laaksonen, J. and Puolam{\"a}ki, K. and others},journal={Virtual reality},volume={15},number={2},pages={161--173},year={2011},publisher={Springer-Verlag},}
2010
MLSP
Contextual information access with augmented reality
A. Ajanki, M. Billinghurst, T. Järvenpää, and 8 more authors
In International Workshop on Machine Learning for Signal Processing 2010
@inproceedings{ajanki2010contextual,title={Contextual information access with augmented reality},author={Ajanki, A. and Billinghurst, M. and J{\"a}rvenp{\"a}{\"a}, T. and Kandemir, M. and Kaski, S. and Koskela, M. and Kurimo, M. and Laaksonen, J. and Puolam{\"a}ki, K. and Ruokolainen, T. and others},booktitle={International Workshop on Machine Learning for Signal Processing},pages={95--100},year={2010},}
Journal
Automatic segmentation of colon glands using object-graphs
C. Gunduz-Demir, M. Kandemir, A.B. Tosun, and 1 more author
@article{gunduz2010automatic,title={Automatic segmentation of colon glands using object-graphs},author={Gunduz-Demir, C. and Kandemir, M. and Tosun, A.B. and Sokmensuer, C.},journal={Medical image analysis},volume={14},number={1},pages={1--12},year={2010},publisher={Elsevier},}
ETRA
Inferring Object Relevance from Gaze in Dynamic Scenes
M. Kandemir, V.M. Saarinen, and S. Kaski
In Symposium on Eye-Tracking Research & Applications 2010
@inproceedings{kandemir2010inferring,title={Inferring Object Relevance from Gaze in Dynamic Scenes},author={Kandemir, M. and Saarinen, V.M. and Kaski, S.},booktitle={Symposium on Eye-Tracking Research \& Applications},pages={105--108},year={2010},}
2009
Journal
Object-oriented texture analysis for the unsupervised segmentation of biopsy images for cancer detection
A.B. Tosun, M. Kandemir, C. Sokmensuer, and 1 more author
@article{tosun2009object,title={Object-oriented texture analysis for the unsupervised segmentation of biopsy images for cancer detection},author={Tosun, A.B. and Kandemir, M. and Sokmensuer, C. and Gunduz-Demir, C.},journal={Pattern Recognition},volume={42},number={6},pages={1104--1112},year={2009},publisher={Elsevier},}
@phdthesis{kandemir2008segmentation,title={Segmentation of Colon Glands by Object Graphs},author={Kandemir, M.},year={2008},school={Bilkent University},}
2007
CGI
A Framework for Real-Time Animation of Liquid-Rigid Body Interaction
@article{kandemir2007framework,title={A Framework for Real-Time Animation of Liquid-Rigid Body Interaction},author={Kandemi̇r, M. and {\c{C}}apin, T. and {\"O}zg{\"u}{\c{c}}, B.},booktitle={Computer Graphics International (CGI)},year={2007},}