A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | Yan | Project Topic | Project Description (tentative) | Reading List | Tentative Concrete Deadlines | what students will get from this project | Member #1 Please sign up | Member #2 Please sign up | ||||||||||||||
2 | ||||||||||||||||||||||
3 | 1 | Space-efficient compound operations for Tensor Networks and Tensorial Neural Networks | https://www.cs.umd.edu/class/fall2018/cmsc498V/slides/TensorBasics.pdf https://arxiv.org/abs/1805.10352 http://www.ece.lsu.edu/jxr/papers-pdf/pldi02.pdf https://arxiv.org/abs/1606.05696 | Tahseen Rabbani | Alex Reustle | |||||||||||||||||
4 | 2 | |||||||||||||||||||||
5 | 3 | Adversarial Example | Ping-yeh Chiang | |||||||||||||||||||
6 | 4 | PAC-RL (tabular) | - Near-optimal regret bounds for reinforcement learning: http://www.jmlr.org/papers/volume11/jaksch10a/jaksch10a.pdf - Sample complexity of episodic fixed-horizon reinforcement learning https://papers.nips.cc/paper/5827-sample-complexity-of-episodic-fixed-horizon-reinforcement-learning.pdf - Minimax regret bounds for reinforcement learning https://arxiv.org/pdf/1703.05449.pdf - Unifying PAC and Regret: Uniform PAC Bounds for Episodic Reinforcement Learning http://papers.nips.cc/paper/7154-unifying-pac-and-regret-uniform-pac-bounds-for-episodic-reinforcement-learning.pdf - Policy Certificates: Towards Accountable Reinforcement Learning https://arxiv.org/pdf/1811.03056.pdf - Optimistic posterior sampling for reinforcement learning: worst-case regret bounds https://papers.nips.cc/paper/6718-optimistic-posterior-sampling-for-reinforcement-learning-worst-case-regret-bounds.pdf - Regret Bounds for Reinforcement Learning via Markov Chain Concentration https://arxiv.org/pdf/1808.01813.pdf - Is Q-learning Provably Efficient? https://arxiv.org/pdf/1807.03765.pdf | Peng Wan | Yanchao Sun | |||||||||||||||||
7 | 5 | Differentially Private Topic Model | ||||||||||||||||||||
8 | 6 | Spectral methods for exploration in POMDPs | - Reinforcement Learning of POMDPs using Spectral Methods http://proceedings.mlr.press/v49/azizzadenesheli16a.pdf - A PAC RL Algorithm for Episodic POMDPs https://arxiv.org/pdf/1605.08062.pdf | Alex Levine | Marina Knittel | |||||||||||||||||
9 | 7 | Generalization of Neural Networks | Sweta Agrawal | Samyadeep | ||||||||||||||||||
10 | 8 | Unsupervised Boosting | Lillian Huang | Jamie Matthews | ||||||||||||||||||
11 | 9 | Off-policy evaluation | - Eligibility traces for off-policy policy evaluation https://scholarworks.umass.edu/cgi/viewcontent.cgi?article=1079&context=cs_faculty_pubs - Doubly Robust Off-policy Value Evaluation for Reinforcement Learning http://nanjiang.cs.illinois.edu/files/ICML2016-DR.pdf - Optimal and Adaptive Off-Policy Evaluation in Contextual Bandits http://proceedings.mlr.press/v70/wang17a/wang17a.pdf - Data-efficient off-policy policy evaluation for reinforcement learning http://proceedings.mlr.press/v48/thomasa16.pdf - Using Options and Covariance Testing for Long Horizon Off-Policy Policy Evaluation http://papers.nips.cc/paper/6843-using-options-and-covariance-testing-for-long-horizon-off-policy-policy-evaluation.pdf - Effective Evaluation Using Logged Bandit Feedback from Multiple Loggers https://arxiv.org/pdf/1703.06180.pdf - Balanced Policy Evaluation and Learning http://papers.nips.cc/paper/8105-balanced-policy-evaluation-and-learning.pdf - Breaking the Curse of Horizon: Infinite-Horizon Off-Policy Estimation http://papers.nips.cc/paper/7781-breaking-the-curse-of-horizon-infinite-horizon-off-policy-estimation.pdf | Zeyad Emam | Daniel Lichy | |||||||||||||||||
12 | 10 | POMDPs and predictive state representations(PSRs) | - Point-based value iteration: An anytime algorithm for POMDPs http://www.fore.robot.cc/papers/Pineau03a.pdf - Perseus: Randomized Point-based Value Iteration for POMDPs - Closing the learning–planning loop with predictive state representations - Efficient Learning and Planning with Compressed Predictive States https://arxiv.org/abs/1312.0286 - Hilbert Space Embeddings of Hidden Markov Models https://www.researchgate.net/profile/Alexander_Smola/publication/221345134_Hilbert_Space_Embeddings_of_Hidden_Markov_Models/links/0912f5105f839ae5a9000000/Hilbert-Space-Embeddings-of-Hidden-Markov-Models.pdf - Low-Rank Spectral Learning with Weighted Loss Functions http://nanjiang.cs.illinois.edu/files/lowrank_aistats15.pdf - Completing State Representations using Spectral Learning http://nanjiang.cs.illinois.edu/files/nips18-psrf.pdf | Goonwanth | ||||||||||||||||||
13 | 11 | Automatic searching of neural network structure | ||||||||||||||||||||
14 | 12 | Defense against Adversarial Examples | Use a provable differentiable lower bound to the closest adversarial example for defense against adversarial examples | https://arxiv.org/abs/1902.01235 | Sahil Singla | |||||||||||||||||
15 | 13 | Bounding Fairness transfer between tasks | Goal: Transfer of Machine Learning Fairness across Tasks. In the context of fairness, we are interested in the setting of a single domain which may have multiple tasks.This might exist in the context of transfer learning or in multi-task learning (MTL). | 1. A Survey of Transfer Learning, Weiss et al. [2016] 2. A Survey of Transfer Learning, Pan and Yang [2010] 3. A Survey on Multi-Task Learning, Zhang and Yang [2017] 4. A Model of Inductive Bias Learning, Baxter [2000] 5. A notion of task relatedness yielding provable multiple-task learning guarantees, Ben-David and Borbely [2008] 6. Algorithm-Dependent Generalization Bounds for Multi-Task Learning, Liu et al. [2017] : 7. A Model of Inductive Bias Learning, Baxter [2000] 8. A framework for learning predictive structures from multiple tasks and unlabeled data, Ando and 9. Bounds for linear multi-task learning, Maurer [2006a]: 10. The Rademacher complexity of linear transformation classes, Maurer [2006b] 11. The Benefit of Multitask Representation Learning, Maurer et al. [2016] 12. Estimating relatedness via data compression, Juba [2006] 13. Equality of opportunity in supervised learning, Hardt et al. [2016] | Samuel Dooley | |||||||||||||||||
16 | 14 | Interpreting Neural Networks using similarity between the layers | Use Canonical Correlation Analysis (used to compute similarity between NN layers) to emperically assess its potential as an alternative to validation set for tuning hyperparameters and architectural settings, investigate the scope of its use in structural search, as a possible metric to explain the 'overparameterization' phenomenon, and model compression. Post empericial investigations, go for theoretical underpinnings of CCA and why it is able to explain learning dynamics or our results. | - SVCCA: Singular Vector Canonical Correlation Analysis for Deep Learning Dynamics and Interpretability https://arxiv.org/pdf/1706.05806.pdf - Insights on representational similarity in neural networks with canonical correlation http://papers.nips.cc/paper/7815-insights-on-representational-similarity-in-neural-networks-with-canonical-correlation.pdf - Understanding Learning Dynamics Of Language Models with SVCCA https://arxiv.org/pdf/1811.00225.pdf - A CLOSER LOOK AT DEEP LEARNING HEURISTICS: LEARNING RATE RESTARTS, WARMUP AND DISTILLATION https://arxiv.org/pdf/1810.13243.pdf - Neural Ordinary Differential Equations https://arxiv.org/pdf/1806.07366.pdf - Towards Understanding the Role of Over-Parametrization in Generalization of Neural Networks https://arxiv.org/pdf/1805.12076.pdf - https://arxiv.org/pdf/1805.10352.pdf | Pranav Goel | Anshul Shah | ||||||||||||||||
17 | 15 | Online Boosting | Leonidas Tsepenekas Eleftheria Briakou | |||||||||||||||||||
18 | 16 | Adversarial Example | -https://arxiv.org/pdf/1801.02610.pdf the first paper is about generating an adversarial example using GAN -https://arxiv.org/pdf/1804.02485.pdf the second paper hinges on the first paper; it tries to generate a fortified neural network that is strong against the adversarial example. -https://arxiv.org/pdf/1812.06570.pdf this is another paper, which uses VAE to defense from adversarial neural network -https://arxiv.org/pdf/1903.00585.pdf similar idea. Using VAE. This one is called PUVAE (purifying VAE). These are easy defenses. -https://arxiv.org/pdf/1806.04646.pdf this paper provides how to attack VAE | Nao Rho | Fan Yang | |||||||||||||||||
19 | 17 | Domain adaptation | http://openaccess.thecvf.com/content_cvpr_2018/papers/Sankaranarayanan_Generate_to_Adapt_CVPR_2018_paper.pdf | Prithviraj Dhar | Shlok Mishra | |||||||||||||||||
20 | ||||||||||||||||||||||
21 | ||||||||||||||||||||||
22 | ||||||||||||||||||||||
23 | ||||||||||||||||||||||
24 | ||||||||||||||||||||||
25 | ||||||||||||||||||||||
26 | ||||||||||||||||||||||
27 | ||||||||||||||||||||||
28 | ||||||||||||||||||||||
29 | ||||||||||||||||||||||
30 | ||||||||||||||||||||||
31 | ||||||||||||||||||||||
32 | ||||||||||||||||||||||
33 | ||||||||||||||||||||||
34 | ||||||||||||||||||||||
35 | ||||||||||||||||||||||
36 | ||||||||||||||||||||||
37 | ||||||||||||||||||||||
38 | ||||||||||||||||||||||
39 | ||||||||||||||||||||||
40 | ||||||||||||||||||||||
41 | ||||||||||||||||||||||
42 | ||||||||||||||||||||||
43 | ||||||||||||||||||||||
44 | ||||||||||||||||||||||
45 | ||||||||||||||||||||||
46 | ||||||||||||||||||||||
47 | ||||||||||||||||||||||
48 | ||||||||||||||||||||||
49 | ||||||||||||||||||||||
50 | ||||||||||||||||||||||
51 | ||||||||||||||||||||||
52 | ||||||||||||||||||||||
53 | ||||||||||||||||||||||
54 | ||||||||||||||||||||||
55 | ||||||||||||||||||||||
56 | ||||||||||||||||||||||
57 | ||||||||||||||||||||||
58 | ||||||||||||||||||||||
59 | ||||||||||||||||||||||
60 | ||||||||||||||||||||||
61 | ||||||||||||||||||||||
62 | ||||||||||||||||||||||
63 | ||||||||||||||||||||||
64 | ||||||||||||||||||||||
65 | ||||||||||||||||||||||
66 | ||||||||||||||||||||||
67 | ||||||||||||||||||||||
68 | ||||||||||||||||||||||
69 | ||||||||||||||||||||||
70 | ||||||||||||||||||||||
71 | ||||||||||||||||||||||
72 | ||||||||||||||||||||||
73 | ||||||||||||||||||||||
74 | ||||||||||||||||||||||
75 | ||||||||||||||||||||||
76 | ||||||||||||||||||||||
77 | ||||||||||||||||||||||
78 | ||||||||||||||||||||||
79 | ||||||||||||||||||||||
80 | ||||||||||||||||||||||
81 | ||||||||||||||||||||||
82 | ||||||||||||||||||||||
83 | ||||||||||||||||||||||
84 | ||||||||||||||||||||||
85 | ||||||||||||||||||||||
86 | ||||||||||||||||||||||
87 | ||||||||||||||||||||||
88 | ||||||||||||||||||||||
89 | ||||||||||||||||||||||
90 | ||||||||||||||||||||||
91 | ||||||||||||||||||||||
92 | ||||||||||||||||||||||
93 | ||||||||||||||||||||||
94 | ||||||||||||||||||||||
95 | ||||||||||||||||||||||
96 | ||||||||||||||||||||||
97 | ||||||||||||||||||||||
98 | ||||||||||||||||||||||
99 | ||||||||||||||||||||||
100 |