AA: Human and Machine decisions

Tutor: Alexander Aushev

How can we improve interaction between people and decision-making AIs? The interdisciplinary field of human-machine interaction attempts to understand the process of human decision-making, builds predictive models for it, and helps people to make good decisions. The need for human-machine interaction arises in behavioural economics, psychology and neuroscience with applications ranging from public policies to daily routines. For this topic, you will review papers related to human-machine interactions, focusing on challenges, applications and existing solutions for improving machine learning models with theories of decision-making and vice versa.

Prerequisite: Basic understanding of reinforcement learning and probabilities.

References:

  • I. Mahmud S, Lin X, Kim JH. Interface for Human Machine Interaction for assistant devices: a review. 2020.
  • II. Young SN, Peschel JM. Review of human–machine interfaces for small unmanned systems with robotic manipulators. 2020.
  • III. Kun AL. Human-machine interaction for vehicles: Review and outlook. 2018.


AD: Collaborative multi-agent DRL

Tutor: Anton Debner

Computer games offer an excellent platform for researching Deep Reinforcement Learning (DRL) methods. In 2015, Mnih et al [0] showed that DRL can be used to play Atari games with human-like performance. In 2019, AlphaStar [1] learned to beat the best StarCraft 2 players and Open AI research showed [2] that DRL agents can learn to collaboratively use unintended game mechanics to complete the task. A collaborative multi-agent system means that there are multiple agents working together to accomplish a common goal. For example, imagine a scenario [3] where two agents are moving furniture from one apartment to another. Each agent is equipped with its own AI, and they are capable of lifting objects on their own. However, lifting a large TV requires collaboration from both agents. They need to communicate their intent to lift, as lifting the TV from one side only may lead to dropping the TV on the floor. With DRL, these agents have to learn not only to lift and carry objects, but also when to communicate, what to communicate and how to react to communication. This topic is mostly a literature survey. It is beneficial to know about deep learning and deep reinforcement learning in advance, but it is not strictly necessary.

References:

  • [0] Mnih, V., Kavukcuoglu, K., Silver, D. et al. Human-level control through deep reinforcement learning. Nature 518, 529–533 (2015). https://doi.org/10.1038/nature14236
  • [1] Vinyals, O., Babuschkin, I., Czarnecki, W.M. et al. Grandmaster level in StarCraft II using multi-agent reinforcement learning. Nature 575, 350–354 (2019). https://doi.org/10.1038/s41586-019-1724-z
  • [2] OpenAI: Emergent Tool Use from Multi-Agent Interaction, https://openai.com/blog/emergent-tool-use/
  • [3] Jain, Unnat, et al. "Two body problem: Collaborative visual task completion." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019. https://arxiv.org/abs/1904.05879


AJ1: Normalizing flows for graph structured data

Tutor: Anirudh Jain

Graph structured data provides useful representation for a lot of domains such as social networks, bioinformatics or robotics. Generative models allow us to learn a latent space from the training data which can be explored for data generation such as for drug discovery with molecular graphs. Graph structure imposes additional challenges due to arbitrary shape and size of graphs and discrete structure. Normalizing flows present a promising direction to map a base distribution to a discrete graph distribution. In this project, you will review the SOTA methods for flow based generative methods and adapt them for graph generation.

References:

  • https://arxiv.org/abs/2001.09382


AJ2: Reinforcement learning for drug discovery

Tutor: Anirudh Jain

Generative models have been proposed to efficiently search the large chemical space for viable drugs. Viable drug candidates need to satisfy certain constraints such as toxicity, stability etc. We propose to study treating the generative models as reinforcement learning agents and help in drug discovery.

References:

  • https://arxiv.org/abs/1806.02473


BL: Black-box adversarial attacks

Tutor: Blerta Lindqvist

Neural network classifiers are susceptible to attacks that cause misclassification. We focus on black-box attacks where attacks cannot perform back-propagation.

References:

  • https://arxiv.org/pdf/1904.02144.pdf
  • https://arxiv.org/pdf/1905.07121.pdf
  • https://dl.acm.org/doi/pdf/10.1145/3394486.3403225
  • https://dl.acm.org/doi/pdf/10.1145/3321707.3321749
  • https://arxiv.org/pdf/1807.04457.pdf


EV1: Cloud and Local Game Streaming

Tutor: Esa Vikberg

Cloud Gaming is a growing market, with still new companies (Amazon, Netflix, Google, Microsoft, Sony, Nvidia) promoting their own services for streaming on-demand games to consumers. Their development is supported by the increasing prevalence of high speed low-latency networking such as fiber optic internet and 5G. Game streaming also allows streaming games from a local server, either to different screens in a house (Nvidia SHIELD, Steam Link, Virtual Desktop, Airlink) or to gaming partners elsewhere (Steam Remote Play). There is room to focus the project on what interests you the most. You can do a review of the available products, focusing on the technical aspects of the services. It is also possible to do a measurement study on things such as latency and the effects of computing resources.

References:

  • https://ieeexplore.ieee.org/document/6574660
  • https://ieeexplore.ieee.org/document/6818918
  • https://aaltodoc.aalto.fi/handle/123456789/19492


EV2: Bitrate Adaptation Algorithms in Multimedia Streaming

Tutor: Esa Vikberg

Current multimedia streaming methods allow adjusting the quality of the stream on-the-fly based on the bandwidth available. In some instances, the stream is transcoded into multiple qualities, allowing the client to pick the quality from multiple options. In others, the stream is sent between two peers, and the quality is adjusted on the server side to match the network conditions.

Study the used adaptation algorithms, focusing on a subset of use-cases such as Video on Demand or live streaming. You also have the possibility of conducting measurements, either on a real system or in a simulated environment, to see how the algorithms fare in different network conditions.

References:

  • https://c3lab.poliba.it/images/6/65/Gcc-analysis.pdf
  • https://ieeexplore.ieee.org/document/7524428
  • https://www.usenix.org/system/files/conference/nsdi15/nsdi15-paper-dong.pdf
  • https://arxiv.org/pdf/2001.02951.pdf


GI: Real time communication protocols in the internet

Tutor: Gazi Illahi

Real time protocols are the enabling technology of  many real-time applications over the internet such as video conferencing, remote desktop environments, cloud gaming and more recently cloud XR. As the application space is evolving, so have the protocols from RTSP and RTMP to WebRTC and SRTP. The students task would be to do an in depth analysis of real time communication protocols, studying factors like,underlying protocols, reliablity, overhead, latency, fairness, congestion control and fairness.

References:

The references are not exhautive and are provided as a starting point. 

  • Schulzrinne, Henning, Stephen Casner, Ron Frederick, and Van Jacobson. "RTP: A transport protocol for real-time applications." (1996). 
  • Sredojev, Branislav, Dragan Samardzija, and Dragan Posarac. "WebRTC technology overview and signaling solution design and implementation." In 2015 38th international convention on information and communication technology, electronics and microelectronics (MIPRO), pp. 1006-1009. IEEE, 2015. 
  • "SRT Allicance", Secure Reliable Transport Protocol, 2018, [online] Available: https://github.com/Haivision/srt/files/2489142/SRT_Protocol_TechnicalOverviewDRAFT 2018-10-17.pdf.


IAAM: Robust Bayesian time series models for forecasting Heteroscedastic data in financial applications.

Tutor: Izhar Asael Alonzo Matamoros

Study time series models such as GARCH and its variants, SVM, and non Stationary Gaussian processes for the study of heteroscedastic time series and applied in financial applications

References:

  • Student-t SVM: https://www.maths.usyd.edu.au/u/jchan/2011CSDA_BivariateSV.pdf
  • Student-t GARCH: https://www.researchgate.net/publication/46434559_Bayesian_Estimation_of_the_GARCH11_Model_with_Student-t_Innovations
  • Non-stationary GP: http://proceedings.mlr.press/v51/heinonen16.pdf


JH: Scheduling for Asymmetric Multicore Processors

Tutor: Jaakko Harjuhahto

Most modern desktop multicore processors are homogenous: all of the processor cores are identical in instruction set architecture and features (i.e. micro-architecture). However, heterogenous or asymmetric processors with multiple different core designs on the same chip offer benefits for specific use cases, such as mobile devices. These heterogenous architectures present the operating system scheduler with more options to exploit the characteristics of the different processor core types to achieve runtime goals, such as energy efficiency [1].

Practical examples of this design paradigm are ARM's big.LITTLE chips with 'big' cores designed for performance and 'LITTLE' cores prioritizing energy efficiency. Intel's upcoming Alder Lake desktop CPUs feature a similar hybrid design with distinct 'Performance' and 'Efficiency' cores on the same chip.

For this topic, the task is to review current literature on heterogenous multicore scheduling. Defining on a more specific research question is recommended. [2] offers a survey on asymmetric multicore processors, from which a sub-section should be chosen as the focus for the seminar paper. Two examples of more specific topics are given below:

 - How does the scheduler profile processes in order to decide which type of core is optimal for them?

 - Compare how two or more scheduling goals (e.g. performance, fairness, energy consumption) are realized in proposed scheduling policies

Having completed a course on operating systems, such as CS-C3140 Operating Systems at Aalto, or taking one during this semester is highly beneficial for this topic.

References:

  • [1] B. Salami, H. Noori and M. Naghibzadeh, Fairness-Aware Energy Efficient Scheduling on Heterogeneous Multi-Core Processors, IEEE Transactions on Computers, 2021, https://doi.org/10.1109/TC.2020.2984607
  • [2] S. Mittal, A Survey of Techniques for Architecting and Managing Asymmetric Multicore Processors, ACM Computing Surveys, 2016, https://doi.org/10.1145/2856125


LN1: The role and impact of security breaches on users' behaviour

Tutor: Leysan Nurgalieva

A security breach is any incident that results in unauthorized access to data. Depending on the scale and media exposure of a breach, it can increase users' security concerns and influence their adoption and usage behaviour of affected systems. The seminar papers on this topic can look at this phenomenon from different angles: from the analysis of the scale vs impact of a breach to identifying categories or patterns of user behaviours reported in related literature.

References:

  • "Concerns with Handling Exposed Data" - https://www.usenix.org/conference/soups2018/presentation/karunakaran
  • “I cannot do anything”: User’s Behavior and Protection Strategy upon Losing, or Identifying Unauthorized Access to Online Account" - https://www.usenix.org/system/files/soups2020_poster_al-ameen.pdf
  • "Concern But No Action: Consumers' Reactions to the Equifax Data Breach" - https://dl.acm.org/doi/abs/10.1145/3170427.3188510


LN2: The motivations/barriers for implementing security design practices in software development

Tutor: Leysan Nurgalieva

It is important to recognise and address potential security threats during the software development process. However, in practice, it is not always happening. Developers might not incorporate security mechanisms into the systems they build, even though they recognise their importance. Papers on this topic could review and analyse the literature on factors that motivate or discourage developers from implementing security mechanisms.

References:

  • "Inside the Organization: Why Privacy and Security Engineering Is a Challenge for Engineers" - https://ieeexplore.ieee.org/abstract/document/8466102
  • "A review on factors influencing implementation of secure software development practices" - https://publications.waset.org/10005706/a-review-on-factors-influencing-implementation-of-secure-software-development-practices
  • "A Survey on Developer-Centred Security" - https://ieeexplore.ieee.org/abstract/document/8802434


LP: Differential Privacy in Practical Implementations

Tutor: Lukas Prediger

Applying machine learning algorithms on data containing senstive personal information carries the risk of memorisation of specific personal information in the trained model. Examination of the model or its prediction outputs can then leak personal information. 

Differential privacy is a mathematical framework which allows to establish strict theoretical bounds on this leakage and has become the de-facto standard in research and practical application.

However, formulations of differential privacy rarely account for the fact that real-world implementations rely on imperfect approximations of ideal math and have to contend with issues such as imperfect sources of randomness and finite-precision number representations.

In this thesis you will review and summarise such problems arising from implementing ideal formulations of differential privacy on real-world computers and possible mitigation strategies.

References:

  • https://link.springer.com/chapter/10.1007/11761679_29
  • https://link.springer.com/chapter/10.1007/978-3-642-03356-8_8
  • https://dl.acm.org/doi/abs/10.1145/2382196.2382264
  • https://www.sciencedirect.com/science/article/pii/S0304397516000268
  • https://www.tau.ac.il/~saharon/BigData2018/privacybook.pdf


MK: A survey of deterministic networking

Tutor: Miika Komu

Survey deterministic networking (DETNET) in academic and standardization literature

References:

  • https://datatracker.ietf.org/doc/html/rfc8557
  • https://datatracker.ietf.org/doc/html/rfc8655
  • https://datatracker.ietf.org/doc/html/draft-ietf-detnet-bounded-latency
  • Last Spring, the course included a seminar paper on real-time networking that focused on Time Sensitive Networking (TSN). The paper is good background reading material.


MS: Infrastructure as Code (IaC) security

Tutor: Mohit Sethi

There are many tools for Infrastructure as Code (IaC). In this seminar paper, the student is expected to explain the concept of IaC, provide an overview of popular tools such as Chef, Ansible, Puppet, SaltStack, Terraform, and explain the security features & potential vulnerabilities of these tools in a DevOps toolchain.

References:

  • Chef: https://www.chef.io/products/chef-infra
  • Ansible: https://www.ansible.com/integrations/infrastructure
  • Puppet: https://puppet.com/
  • Terraform: https://www.terraform.io/


NHK: Understanding the Expressivity of Neural Network Control Policies in Reinforcement Learning

Tutor: Nam Hee Kim

Using multi-layered neural network policies is the de facto standard practice for learning-based motion control solutions. However, the relationship between the architecture of neural network backbones of control policies and their performance is poorly understood both theoretically and empirically. The student will focus on (1) designing toy problems for simple reinforcement learning setup, (2) exploring visualization techniques for understanding control granularity within state space, and (3) exploring architectures and hyperparameters of a neural network model while taking sensitivity to randomization into account. This work aims to be a hands-on introduction to deep reinforcement learning (DRL) research where the student will have an opportunity to implement state-of-the-art algorithms as well as examine various numerical footprints of neural network models and their resulting trajectories.

Prerequisite: Familiar with Python for differentiable programming (e.g. PyTorch) and Pythonic ways to do object-oriented programming. Visualization techniques (e.g. Matplotlib). Experiment tracking (e.g. Weights & Biases, Neptune) and managing cluster compute jobs (e.g. SLURM).

Also available as a CS-E4875 topic.

References:

  • Raghu, M., Poole, B., Kleinberg, J., Ganguli, S. & Sohl-Dickstein, J.. (2017). On the Expressive Power of Deep Neural Networks. Proceedings of the 34th International Conference on Machine Learning, in Proceedings of Machine Learning Research 70:2847-2854 Available from https://proceedings.mlr.press/v70/raghu17a.html.
  • Hanin, B. & Rolnick, D.. (2019). Complexity of Linear Regions in Deep Networks. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:2596-2604 Available from https://proceedings.mlr.press/v97/hanin19a.html.
  • Hanin, Boris, and David Rolnick. "Deep relu networks have surprisingly few activation patterns." (2019)
  • Zhang, Xiao, and Dongrui Wu. "Empirical studies on the properties of linear regions in deep neural networks." arXiv preprint arXiv:2001.01072 (2020).


PH: Neural temporal point process

Tutor: Pashupati Hegde

Temporal point processes (TPPs) can be used to model discrete events occurring over continuous-time horizons. Compared to the classical TPP models, which are usually defined to handle simple event patterns, the recent works attempt to define more expressive models by combining the ideas from deep learning. This project on so-called "Neural temporal point processes" will involve surveying the current literature, discussing main design choices, and providing a comparative study of related works. Optionally, one can also perform experiments on small datasets using existing methods and their implementations.

Prerequisites: Basic understanding of probability and statistics, machine learning and programming skills

References:

  • [1] Hongyuan Mei and Jason Eisner. The neural Hawkes process: A neurally self-modulating multivariate point process. NeurIPS, 2017
  • [2] Takahiro Omi, Naonori Ueda, and Kazuyuki Aihara. Fully neural  network based model for general temporal point processes. NeurIPS, 2019.
  • [3] Ricky TQ Chen, Brandon Amos, and Maximilian Nickel. Neural spatio-temporal point processes. ICLR, 2021.


SDP: Computationally modelling and supplementing human decision making.

Tutor: Sebastiaan De Peuter

Recent work in Cognitive Science has shown that human decision making matches closely to planning using a decision tree. Under this model the decision tree is expanded using an optimal expansion policy (optimal given the specific decision problem), and a stopping rule is used to determine when to stop expansion. Once tree expansion stops the action that is best (under the decision tree) is taken. In this project you will implement this model on a specific decision problem, namely trip design. Time permitting we will then expand this model so that it can take into account action recommendations from outside (for example from an assistant/advisor). We will then test, using previously developed AI assistants, whether action recommendations can improve this model's decisions making. This is a theoretical project, there are no human studies planned. Basic knowledge of planning/RL is recommended.

References:

  • https://psyarxiv.com/byaqd/
  • https://www.is.mpg.de/publications/jain2021computational


SR: User-Provided-Fog-Infrastructure:  A game-theoretic approach

Tutor: Sara Ranjbaran

In this project we are going to evaluate the interaction between IoT smart devices and could/fog providers; evaluating how providers compete/cooperate with each other and how it can affect user/consumer behaviour. The goal is to create on-promise IoT cloud points as an extra source of edge services, where IoT devices can request based on their needs and preferences.

References:

  • Iosifidis, George, et al. "Incentive Schemes for User‐Provided Fog Infrastructure." Fog and Fogonomics: Challenges and Practices of Fog Computing, Communication, Networking, Strategy, and Economics (2020): 129-150.
  • Siew, Marie, et al. "Dynamic pricing for resource-quota sharing in multi-access edge computing." IEEE Transactions on Network Science and Engineering 7.4 (2020): 2901-2912.
  • Zavodovski, Aleksandr, et al. "DeCloud: Truthful decentralized double auction for edge clouds." 2019 IEEE 39th International Conference on Distributed Computing Systems (ICDCS). IEEE, 2019.


SS1: Biometric authentication beyond fingers and eyes

Tutor: Sanna Suoranta

First biometric method for authentication that comes to mind is usually based on fingerprints or iris of an eye. However, there are lot of other biometric methods that can be used in authentication. What of them would be most handy to use with computer or/and mobile device and most accepteble for every day use or some specific use?

References:

Some random references to start with:

  • Yuxin Chen, Zhuolin Yang, Ruben Abbou, Pedro Lopes, Ben Y. Zhao, and Haitao Zheng. User Authentication via Electrical Muscle Stimulation. CHI'21. https://doi.org/10.1145/3411764.3445441
  • Zhang Rui and Zheng Yan. A Survey on Biometric Authentication: Toward Secure and Privacy-Preserving Identification. IEEE Access, vol 7, 2018. DOI: 10.1109/ACCESS.2018.2889996


SS2: Continuous Behavioral Authentication

Tutor: Sanna Suoranta

Some techniques allow user authentication as byproduct of common usage of a device. These technologies can be based on e.g. keystrokes, swipe gestures or walking habits. The aim of this article is to tell the current state-of-the-art of continuous authentication for a specific method for a specific use case, chosen by the student (and tutor).

References:

Some random references to start with:

  • Soumik Mondal and Patrick Bours. A study on continuous authentication using a combination of keystroke and mouse biometrics. Neurocomputing, vol 230, 22 March 2017, pages 1-22. https://doi.org/10.1016/j.neucom.2016.11.031
  • Ioannis C. Styios, Olga Thanou, Iosif Androulidakis and Elena Zaitseva. A Review of Continuous Authentication Using Behavioral Biometrics. SEEDA-CECNSM '16: Proceedings of the SouthEast European Design Automation, Computer Engineering, Computer Networks and Social Media ConferenceSeptember 2016 Pages 72–79. https://doi.org/10.1145/2984393.2984403


SS3: Split learning as an alternative to federated learning

Tutor: Stephan Sigg

Split learning has been proposed in 2018 as an alternative to Federated learning, where a server utilizes deep learning models for training or inference without accessing raw data from clients. 

An example configuration could be clients that train a partial deep network up to a specific layer and sends the outputs at that layer to another entity which continues further part of the training without access to raw data. 

In this seminar topic, the student is expected to produce a concise, algorithmical introduction into the concept of split learning. Further, the student shall compare the concept to the concept of federated learning with respect to different performance metrics.

References:

  • https://www.media.mit.edu/projects/distributed-learning-and-collaborative-learning-1/overview/


SS4: CNN over unordered sets

Tutor: Stephan Sigg

Traditional convolutional neural networks assume an implicit ordering over their inputs. This seminar topic is to study and report about approaches to apply CNN structures to unordered sets of points.

Point clouds are often referred to unordered sets of points and mark one important class of unordered inputs. However, point clouds still inherit an order through their 3D coordinates. 

The student is to present a concise presentation on point-cloud processing approaches (e.g. point-based, voxel-based, graph-based) and in addition to conceptionally discuss CNNs for 'bag of ...' type of inputs.

References:

  • https://arxiv.org/pdf/1905.08705


VH: Co-inference techniques for Edge and Fog computing

Tutor: Vesa Hirvisalo

Edge and Fog computing [1] are emerging paradigms for organizing computing services for IoT (Internet of Things). They extend the concepts and practices of cloud computing toward the rapidly increasing number of connected devices. Many aspects of both edge and fog computing are currently under intense research. One of these is utilization of artificial intelligence based on deep learning the IoT applications. However, the edge and fog computing domains have their inherent requirements and restrictions, which calls for efficient inference methods to be used. Various co-inference methods (e.g., [2],[3]) are one approach to this.

The task to is make an overview of co-inference techniques in Edge and Fog computing.

References:

  • [1] Yousefpour & al. All One Needs to Know about Fog Computing and Related Edge Computing Paradigms - A Complete Survey. Journal of Systems Architecture. DOI:10.1016/j.sysarc.2019.02.009
  • [2] S. Teerapittayanon, B. McDanel and H. T. Kung, BranchyNet: Fast inference via early exiting from deep neural networks, 2016 23rd International Conference on Pattern Recognition (ICPR), 2016, pp. 2464-2469, doi: 10.1109/ICPR.2016.7900006.
  • [3] J. Shao and J. Zhang, "BottleNet++: An End-to-End Approach for Feature Compression in Device-Edge Co-Inference Sstems, 2020 IEEE International Conference on Communications Workshops (ICC Workshops), 2020, pp. 1-6, doi: 10.1109/ICCWorkshops49005.2020.9145068.


VTB1: How to improve ADR?

Tutor: Verónica Toro-Betancur

LoRa is a very popular wireless communication technology that has been widely deployed in the world. LoRaWAN is the specification for LoRa networks that defines the Adaptive Data Rate (ADR) mechanism for assigning the communication parameters, to the LoRa devices, that increase the network capacity and decrease the energy consumption. This mechanism is widely used in real-world networks, however, it is known that such networks do not achieve satisfactory performance under dynamic scenarios. For instance, under wireless channels with a high variance.

The goal of this topic is to review the many variants to ADR that have been proposed in the literature. Moreover, the student should investigate the dependence on the channel variance of the constants used in the ADR algorithm. For the latter task, a Python LoRa simulator with and ADR implementation will be provided.

References:

  • Kim, D. Y., Kim, S., Hassan, H., & Park, J. H. (2017). Adaptive data rate control in low power wide area networks for long range IoT services. Journal of computational science, 22, 171-178.
  • Slabicki, M., Premsankar, G., & Di Francesco, M. (2018, April). Adaptive configuration of LoRa networks for dense IoT deployments. In NOMS 2018-2018 IEEE/IFIP Network Operations and Management Symposium (pp. 1-9). IEEE.
  • Benkahla, N., Tounsi, H., Ye-Qiong, S. O. N. G., & Frikha, M. (2019, June). Enhanced ADR for LoRaWAN networks with mobility. In 2019 15th International Wireless Communications & Mobile Computing Conference (IWCMC) (pp. 1-6). IEEE.
  • Kufakunesu, R., Hancke, G. P., & Abu-Mahfouz, A. M. (2020). A survey on Adaptive Data Rate optimization in LoRaWAN: Recent solutions and major challenges. Sensors, 20(18), 5044.


VTB2: Modeling CSMA/CA

Tutor: Verónica Toro-Betancur

A channel access method determines the transmission dynamics in a wireless network. That is, it defines the steps that a device has to take before transmitting a data packet, e.g., listening to the channel first and, if it is idle, transmit the packet. These methods aim to minimize the collisions between packets in the network. Several channel access methods have been proposed and adopted in real-world deployments. One of the most popular methods is CSMA/CA (Carrier-sense multiple access with collision avoidance) as it is the default method used by WiFi and ZigBee. Research involving any of these technologies benefit from a complete model of the CSMA/CA technique.

The goal of this topic is to review existing CSMA/CA models and highlight their complexity and accuracy. Students interested in this topic should be willing to dig into (sometimes) complex mathematical formulations.

References:

  • Xin Wang and K. Kar, "Throughput modelling and fairness issues in CSMA/CA based ad-hoc networks," Proceedings IEEE 24th Annual Joint Conference of the IEEE Computer and Communications Societies., 2005, pp. 23-34 vol. 1, doi: 10.1109/INFCOM.2005.1497875.
  • Gamal, M., Sadek, N., Rizk, M. R., & Ahmed, M. A. E. (2020). Optimization and modeling of modified unslotted CSMA/CA for wireless sensor networks. Alexandria Engineering Journal, 59(2), 681-691.
  • Peregudov, M. A. E., Steshkovoy, A. S., & Boyko, A. A. (2018). Probabilistic random multiple access procedure model to the CSMA/CA type medium. Informatics and Automation, 59, 92-114.
  • Busson, A., & Chelius, G. (2009, October). Point processes for interference modeling in csma/ca ad-hoc networks. In Proceedings of the 6th ACM symposium on Performance evaluation of wireless ad hoc, sensor, and ubiquitous networks (pp. 33-40).
  • Wang, F., Li, D., & Zhao, Y. (2011). Analysis of csma/ca in IEEE 802.15. 4. IET communications, 5(15), 2187-2195.


WM: Reinforcement learning-based resource allocation

Tutor: Wencan Mao

Recently, reinforcement learning-based algorithms are used for computation offloading and task allocation in complex vehicular networks. The survey aims to compare these works in terms of latency, quality, computational accuracy, time complexity, etc.

References:

  • C. Zhu, Y. -H. Chiang, Y. Xiao and Y. Ji, "FlexSensing: A QoI and Latency-Aware Task Allocation Scheme for Vehicle-Based Visual Crowdsourcing via Deep Q-Network," in IEEE Internet of Things Journal, vol. 8, no. 9, pp. 7625-7637, 1 May1, 2021, doi: 10.1109/JIOT.2020.3040615.
  • J. Wang, C. Jiang, K. Zhang, T. Q. S. Quek, Y. Ren and L. Hanzo, "Vehicular Sensing Networks in a Smart City: Principles, Technologies and Applications," in IEEE Wireless Communications, vol. 25, no. 1, pp. 122-132, February 2018, doi: 10.1109/MWC.2017.1600275.
  • J. Shi, J. Du, J. Wang, J. Wang and J. Yuan, "Priority-Aware Task Offloading in Vehicular Fog Computing Based on Deep Reinforcement Learning," in IEEE Transactions on Vehicular Technology, vol. 69, no. 12, pp. 16067-16081, Dec. 2020, doi: 10.1109/TVT.2020.3041929.


ZS: A study of particle-based variational inference for Bayesian machine learning

Tutor: Zheyang Shen

Bayesian inference of intractable distributions is a cornerstone of probabilistic machine learning. In this seminar project, we explore particle-based variational inference (ParVI), a novel family of quasi Monte Carlo method that draws approximate samples from an un-normalized target distribution by incrementally transporting a set of particle samples towards the target distribution [1]. For this study we will start by investigating the theoretical properties of different ParVI algorithms (e.g., [2]) as vector fields in the Wasserstein space, and conduct simple experiments on the practical applicability of such algorithms. It is preferred that you have rudimentary understanding of Bayesian inference, probability theory and programming.

References:

  • [1] Liu, Q., & Wang, D. (2016). Stein Variational Gradient Descent: A General Purpose Bayesian Inference Algorithm. In Advances in Neural Information Processing Systems (Vol. 29, pp. 2378–2386).
  • [2] Liu, C., Zhuo, J., Cheng, P., Zhang, R., & Zhu, J. (2019). Understanding and Accelerating Particle-Based Variational Inference. In International Conference on Machine Learning (pp. 4082–4092).


Viimeksi muutettu: torstaina 16. syyskuuta 2021, 12.58