\\ \\ \\ \\ ====== Lectures ====== ^ date ^ week ^ topic ^ slides + videos ^ | 19.09.2022 | 1 | Lec 1: __Machine learning 101:__ model, loss, learning, issues, regression, classification | {{ :courses:b3b33vir:lectures:01_intro_learning_regression_classification.pdf |}}\\ [[https://www.youtube.com/watch?v=rzar6Uz9Ri0&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=1| lec1_recording]]| | 26.09.2022 | 2 | Lec 2: __Under the hood of a linear classifier:__ two-class and multi-class linear classifier on RGB images | {{ :courses:b3b33vir:lectures:02_classification.pdf |}} \\ [[https://www.youtube.com/watch?v=bfdqV94KvbI&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=2|lec2_recording]] | | 03.10.2022 | 3 | Lec 3: __Where the hell does the loss come from?__ MAP and ML estimate, KL divergence and losses. | {{ :courses:b3b33vir:lectures:03_mle.pdf |}} \\ [[https://www.youtube.com/watch?v=aSVsud-bRqg&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8| lec3_recording]] | | 10.10.2022 | 4 | Lec 4: __The story of the cat's brain surgery:__ fully-connected NN + fast backpropagation via Vector-Jacobian-Product (VJP), cortex + convolutional layer | \\ {{ :courses:b3b33vir:lectures:04b_convnets.pdf |}} \\ [[https://www.youtube.com/watch?v=hLjWS5C1IZ4&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=4|lec4_recording]] | | 17.10.2022 | 5 | Lec 5: __ Under the hood of auto-differentiation:__ Vector-Jacobian-Product (VJP) vs chainrule and multiplication of Jacobians, convolutional layer and its VJP | {{ :courses:b3b33vir:lectures:04a_fully_connected_neural_nets.pdf |}} \\ [[https://www.youtube.com/watch?v=kTCTrMrdH4o&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=5|lec5_recording]] | | 24.10.2022 | 6 | **Midterm test** | {{ :courses:b3b33vir:lectures:vir_2022_training_questions_midterm_test.pdf |}}\\ ("pure SGD" = "SGD", i.e. momentum = 0) | | 31.10.2022 | 7 | Lec 6: __Why is learning prone to fail? - Structural issues:__ layers + issues, batch-norm, drop-out | {{ :courses:b3b33vir:lectures:layers.pdf |}} \\ [[https://www.youtube.com/watch?v=GuHPjOrZPsk&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=6|lec6_recording]] | | 07.11.2022 | 8 | Lec 7: __Why is learning prone to fail? - Optim. issues:__ optimization vs learning, KL divergence, SGD, momentum, convergence rate, Adagrad, RMSProp, AdamOptimizer, diminishing/exploding gradient, oscillation, double descent | {{ :courses:b3b33vir:lectures:training.pdf |}} \\ [[https://www.youtube.com/watch?v=7MaNFClK-Mw&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=7|lec7_recording]] | | 14.11.2022 | 9 | Lec 8: __What can('t) we do with a deep net?:__ Classification (ResNet, Squeeze and Excitation Nets), Segmentation (DeepLab), Detection (Yolo, fast-RCNN), Regression (OpenPose), Spatial Transformer Nets, Memory and attention (recurrent nets, Image transformers with attention module)| {{ :courses:b3b33vir:lectures:08a_architectures_class_segm_simplified.pdf |}} \\ {{ :courses:b3b33vir:lectures:08b_architectures_simple_reg_det_stn.pdf |}} \\ [[https://www.youtube.com/watch?v=y3BcoW1pWuc&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=8|lec8_recording]] | | 21.11.2022 | 10 | Lec 9: __Reinforcement learning:__ Approximated Q-learning, DQN, DDPG, Derivation of the policy gradient (REINFORCE), A2C, Inverse RL, Applications, \\ \\ Lec 9 3/4:__Learning to optimize:__ Backpropagation through unconstrained and constrained optimization problems, application (end-to-end differentiable modules cvxpy, gradSLAM, gradMPC, gradODE, pytorch3d) \\ \\ Lec 9 9/10:__Self-supervision__: (pseudo-labeling, privileged information, monodepth)| {{ :courses:b3b33vir:lectures:09_reinforcement_learning.pdf |}} \\ [[https://www.youtube.com/watch?v=QMEb03GduBg&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=9|lec9_recording]] | | 28.11.2022 | 11 | Lec 10: __Generative Adversial Networks:__ Guest lecture by David Coufal, ÚI AV ČR | [[https://www.youtube.com/watch?v=xQC0kRf_QvY&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=10| lec10_recording]] \\ {{ :courses:b3b33vir:lectures:vir2022_a.pdf |}}| | 05.12.2022 | 12 | Lec 11: __Normalizing Flows:__ Guest lecture by David Coufal, ÚI AV ČR| [[https://www.youtube.com/watch?v=TZgFTUrB694&list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8&index=11|lec11_recording]] \\ {{ :courses:b3b33vir:lectures:vir2022_b.pdf |}}| | 12.12.2022 | 13 | **Exam test** | {{ :courses:b3b33vir:lectures:vir_2022_training_questions_exam_test.pdf |}} | | 09.01.2023 | 15 | Lec 12: __Deep learning for satellite imagery:__ Guest lecture of [[https://sites.google.com/site/reinsmic/|Michal Reinstein]] (chief scientist in [[https://spaceknow.com/| Spaceknow]]) | | The playlist of all lecture recordings is [[https://www.youtube.com/playlist?list=PLQL6z4JeTTQkQ7j33fSemZ7uokvvyjBQ8|here]]. \\ \\ \\ ---- ====== Lectures 2021 ====== ^ date ^ week ^ topic ^ slides + videos ^ | 20.09.2021 | 1 | Lec 1: __MLE regression:__ derivation of L2 loss, prior | [[https://drive.google.com/drive/folders/17rGQ9t1alEI2N87fLntJWQr0SKZkzCOx?usp=sharing|lec_01_MLE_regression]] | | 27.09.2021 | 2 | Lec 2: __MLE classifier:__ derivation of cross-entropy loss and logistic loss and linear classifier | {{ :courses:b3b33vir:lectures:mle_02_classification.pdf | lec_02_MLE_classifiction(pdf)}} \\ [[https://www.youtube.com/watch?v=dM_sOifbRfk&list=PLgm4EG9KbqGLJ0ykHLTHyhui1f0d7vb2v|lec_02_recording]]\\ {{ :courses:b3b33vir:lectures:mle_linear_classifier_00.py.zip | codes}} | | 04.10.2021 | 3 | Lec 3: __Fully-connected network:__ computational graph + backpropagation | {{ :courses:b3b33vir:lectures:neural_nets.pdf | lec_03_NN(pdf)}} \\ [[https://www.youtube.com/watch?v=u1wJx9_VeJI&list=PLgm4EG9KbqGLJ0ykHLTHyhui1f0d7vb2v&index=2|lec_03_recording]]| | 11.10.2021 | 4 | Lec 4: __ConvNet:__ convolutional layer + backpropagation | {{:courses:b3b33vir:lectures:convnets.pdf | lec_04_ConvNets(pdf)}} \\ [[https://www.youtube.com/watch?v=6QCA6shl9Ew&list=PLgm4EG9KbqGLJ0ykHLTHyhui1f0d7vb2v&index=3 | lec_04_recording]]| | 18.10.2021 | 5 | Lec 5: __Training:__ SGD, momentum, convergence rate, Adagrad, RMSProp, AdamOptimizer, diminishing/exploding gradient, oscilation | {{ :courses:b3b33vir:lectures:training.pdf |lec_05_training(pdf)}} \\ [[https://www.youtube.com/watch?v=iCkWyDNWgbo&list=PLgm4EG9KbqGLJ0ykHLTHyhui1f0d7vb2v&index=4|lec_05_recording]] | | 25.10.2021 | 6 | Lec 6: __Layers:__ Convolution, Activation functions, Batch-Instance Norm, MaxPooling, Losses + backpropagation \\ **Test T1** | {{ :courses:b3b33vir:lectures:convnets.pdf | lec_06_ConvNets}} \\ {{ :courses:b3b33vir:lectures:training_examples.zip | training_examples.zip}}| | 01.11.2021 | 7 | Lec 7: __Architectures I:__ classification (ResNet, Squeeze and `excitation Nets)| {{ :courses:b3b33vir:lectures:architectures_classification_segmentation.pdf | lec_07_classification_segmentation(pdf)}} \\ [[https://www.youtube.com/watch?v=N0cwYNDDncg| lec_07_recording]] | | 08.11.2021 | 8 | Lec 8: __Architectures II:__ segmentation (DeepLab), pose regression (OpenPose), | {{ :courses:b3b33vir:lectures:architectures_classification_segmentation.pdf | lec_08_classification_segmentation(pdf)}} \\ [[https://www.youtube.com/watch?v=JocIqJriQ9w&list=PLgm4EG9KbqGLJ0ykHLTHyhui1f0d7vb2v&index=5| lec_08_recording]] | | 15.11.2021 | 9 | Lec 9: __Architectures III:__ detection (Yolo), depth regression, spatial transformer nets | {{ :courses:b3b33vir:lectures:architectures_pose_regression_detection_spatial_nets.pdf |lec_09_regression_detection_stn(pdf)}} \\ [[https://www.youtube.com/playlist?list=PLQL6z4JeTTQnv27IWAY6NLafP6xiflmHe| SVTI streams and recordings]] | | 22.11.2021 | 10 | Lec 10: __Reinforcement Learning I:__ Approximated Q-learning, DQN, DDPG, Policy gradient (REINFORCE), | {{ :courses:b3b33vir:lectures:reinforcement_learning.pdf | lec_10_DQN_DDPG_REINFORCE(pdf)}} \\ [[https://www.youtube.com/watch?v=ujXiOOT4fIY&list=PLQL6z4JeTTQkJkapNvV1f564fdFRRSSvn&index=1&t=350s | lec_10_recording]] | | 29.11.2021 | 11 | Lec 11: __Generative Adversarial Networks:__ \\ guest lecture by David Coufal, ÚI AV ČR) | {{ :courses:b3b33vir:lectures:coufal_gan.pdf |lect_11_GANs(pdf)}} \\ [[https://www.youtube.com/watch?v=XxcEn_p9hPo&list=PLQL6z4JeTTQkJkapNvV1f564fdFRRSSvn&index=2 | lec_11_recording]] | | 06.12.2021 | 12 | Lec 12: __Reinforcement Learning II:__ Derivation of the policy gradient, A2C, Inverse Reinforcement Learning, Applications, end-to-end differentiable modules (cvxpy, gradSLAM, pytorch3d) | {{ :courses:b3b33vir:lectures:reinforcement_learning.pdf | lec_12_A2C_IRL(pdf)}}\\ [[https://www.youtube.com/watch?v=uihS_Ft3PcQ&list=PLQL6z4JeTTQkJkapNvV1f564fdFRRSSvn&index=3|lec_12_recording]] | | 13.12.2021 | 13 | Lec 13: __Memory and attention:__ Recurrent nets, Image transformers with attention module, Depth completition (monodepth), Graph convolution, | {{ :courses:b3b33vir:lectures:memory_attention.pdf | lec_13_memory_attention(pdf)}} \\ [[https://www.youtube.com/watch?v=6aIbiP6muN0&list=PLQL6z4JeTTQkJkapNvV1f564fdFRRSSvn&index=4|lec_13_recording]] | | 03.01.2022 | 15 | Lec 14: **Exam Test ET** | | [[https://www.youtube.com/playlist?list=PLQL6z4JeTTQnv27IWAY6NLafP6xiflmHe| All SVTI streams]] \\ \\ \\ ---- ====== Lectures 2020 ====== ^ date ^ week ^ topic ^ slides+videos+codes ^ | 21.09.2020 | 1 | Lec 1: __Regression/Classification as MLE:__ derivation of L2-loss, cross-entropy loss, logistic loss \\ \\ Lec 2: __Neural networks:__ Fully-connected layer + computational graph + backpropagation |[[https://bbb04.felk.cvut.cz/playback/presentation/2.0/playback.html?meetingId=d22e1860a2480c7f10b5ff3e7889e6aed0d0e089-1600689601426| lec_1_video]] (incorrect audio sync!) {{ :courses:b3b33vir:vir_outline.pdf |}}{{ :courses:b3b33vir:mle_01_regression.pdf |}} {{ :courses:b3b33vir:mle_regression.py |}}\\ \\ [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=80ecdf8cd9a3fa2fe2b327ac2c337c3f3d9bd74c-1600758001518|lec2_internal_viewer]][[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=80ecdf8cd9a3fa2fe2b327ac2c337c3f3d9bd74c-1600758001518|lec_2_video]] {{ :courses:b3b33vir:mle_02_classification.pdf |}} {{ :courses:b3b33vir:mle_linear_classifier.py |}}| | 28.9.2020 | 2 | //State holidays// | | | 05.10.2020 | 3 | Lec 3: __ConvNet__: convolutional layer + backpropagation | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=ef299fc9992cf65cfd7450abc6611ba2ea765947-1601900101765|lec3_internal_viewer]] [[https://bbb04.felk.cvut.cz//presentation/ef299fc9992cf65cfd7450abc6611ba2ea765947-1601900101765/ef299fc9992cf65cfd7450abc6611ba2ea765947-1601900101765.mp4|lec3.video]] {{ :courses:b3b33vir:lectures:neural_nets.pdf |}}, {{ :courses:b3b33vir:lectures:convnets.pdf |}} {{ :courses:b3b33vir:lectures:convolution_codes.zip |}}| | 12.10.2020 | 4 | Lec 4: __Training:__ SGD, momentum, convergence rate, Adagrad, RMSProp, AdamOptimizer, diminishing/exploding gradient, oscilation | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=fef9b28b5108e29651ee7dc544e254ac67c0e19e-1602504602053|lec4_internal_viewer]] [[https://bbb04.felk.cvut.cz//presentation/fef9b28b5108e29651ee7dc544e254ac67c0e19e-1602504602053/fef9b28b5108e29651ee7dc544e254ac67c0e19e-1602504602053.mp4|lec4_video]] \\ {{ :courses:b3b33vir:lectures:training.pdf | training.pdf (corrected slides)}} | | 19.10.2020 | 5 | Lec 5: __Layers:__ Convolution, Activation functions, Batch-Instance Norm, MaxPooling, Losses + backpropagation | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=465d9eca29083a573789141cebd81a390d8d2804-1603109701558|lec5_internal_viewer]],\\ {{ :courses:b3b33vir:lectures:layers.pdf | layers.pdf (corrected)}} | | 26.10.2020 | 6 | Lec 6: __Architectures I:__ classification (ResNet), segmentation (DeepLab) \\ **Test T1** | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=b918d7ab577b5bdb7488fcfce6400228a29d22b4-1603720801666|lec6_internal_viewer]] \\ {{ :courses:b3b33vir:lectures:architectures_i.pdf |}} | | 02.11.2020 | 7 | Lec 7: __Learning from unlabelled data:__ Self-supervision: Contrastive learning, rotation, jigsaw, Colorization\\ Weak-supervision: Multiple-instance learning, physical constraints; guest lecture by Patrik Vacek | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=b2b062d1a9548833b01bcb67d699f28ae07b0231-1604322901782|lec7_internal_viewer]] \\ {{ :courses:b3b33vir:lectures:unlabelled.pdf |unlabelled_data.pdf}} | | 09.11.2020 | 8 | Lec 8: __Architectures II:__ pose regression, detection (Yolo), depth regression, spatial transformer nets, LIFTs | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=12b2dabcd639b64194c0114da8dabc8566c8946e-1604927701276|lec8_internal_viewer]] \\ {{ :courses:b3b33vir:lectures:architectures_ii.pdf |}} | | 16.11.2020 | 9 | Lec 9: __Structured inputs: Recurrent neural networks, Convolution in 1,2 and 3D and other structures:__ \\ guest lecture by Teymur Azayev | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=54af05d6433ee1861948ae6a090d67ebab0467aa-1605532502084|lec9_internal_viewer]] \\ {{ :courses:b3b33vir:lectures:lecture_9_structure.pdf |}} | | 23.11.2020 | 10 | Lec 10: __Reinforcement Learning I:__ DQN, GAE+TD(lambda), DDPG | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=6672dd22edc41171667ab697eaaddabcfcfd4774-1606137301264 | lec10_internal_viewer]] \\ {{ :courses:b3b33vir:lectures:reinforcement_learning.pdf |}} | | 30.11.2020 | 11 | Lec 11: __Reinforcement Learning II:__ Policy gradients (REINFORCE), Inverse Reinforcement Learning, Applications, \\ \\ __What did not fit to lectures:__ domain transfer, MAML, monodepth, depth from symmetries, Pytorch3D, CvxPyLayer (it did not fit even here ;-)) | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=3348449de859d49e8ac58ae950252f998e94e8de-1606742580247 | lec11_internal_viewer]] \\ {{ :courses:b3b33vir:lectures:reinforcement_learning.pdf |}} | | 07.12.2020 | 12 | Lec 12: __Generative Adversarial Networks:__ \\ guest lecture by David Coufal, ÚI AV ČR) | [[https://bbb04.felk.cvut.cz//playback/presentation/2.0/playback.html?meetingId=56bde53329b8ea6adc4fd99c6b7f16ce61fcd378-1607347623288 | lec12_internal_viewer]] \\ {{ :courses:b3b33vir:vir2019.pdf | gans_coufal_vir_2019}} | | 14.12.2020 | 13 | **Exam Test ET** | see BRUTE for results | | 04.01.2021 | 14 | Selected teams present their semestral works (the rest presents during labs) | |