| 1. |
- Isaksson, Martin, et al.
(author)
-
Adaptive Expert Models for Federated Learning
- 2023
-
In: <em>Lecture Notes in Computer Science </em>Volume 13448 Pages 1 - 16 2023. - Cham : Springer Science and Business Media Deutschland GmbH. - 9783031289958 ; 13448 LNAI, s. 1-16
-
Conference paper (peer-reviewed)abstract
- Federated Learning (FL) is a promising framework for distributed learning when data is private and sensitive. However, the state-of-the-art solutions in this framework are not optimal when data is heterogeneous and non-IID. We propose a practical and robust approach to personalization in FL that adjusts to heterogeneous and non-IID data by balancing exploration and exploitation of several global models. To achieve our aim of personalization, we use a Mixture of Experts (MoE) that learns to group clients that are similar to each other, while using the global models more efficiently. We show that our approach achieves an accuracy up to 29.78% better than the state-of-the-art and up to 4.38% better compared to a local model in a pathological non-IID setting, even though we tune our approach in the IID setting. © 2023, The Author(s)
|
|
| 2. |
- Lidstrom, D, et al.
(author)
-
Agent based match racing simulations : Starting practice
- 2022
-
In: SNAME 24th Chesapeake Sailing Yacht Symposium, CSYS 2022. - : Society of Naval Architects and Marine Engineers.
-
Conference paper (peer-reviewed)abstract
- Match racing starts in sailing are strategically complex and of great importance for the outcome of a race. With the return of the America's Cup to upwind starts and the World Match Racing Tour attracting young and development sailors, the tactical skills necessary to master the starts could be trained and learned by means of computer simulations to assess a large range of approaches to the starting box. This project used game theory to model the start of a match race, intending to develop and study strategies using Monte-Carlo tree search to estimate the utility of a player's potential moves throughout a race. Strategies that utilised the utility estimated in different ways were defined and tested against each other through means of simulation and with an expert advice on match racing start strategy from a sailor's perspective. The results show that the strategies that put greater emphasis on what the opponent might do, perform better than those that did not. It is concluded that Monte-Carlo tree search can provide a basis for decision making in match races and that it has potential for further use.
|
|
| 3. |
- Håkansson, Maria, et al.
(author)
-
Facilitating Mobile Music Sharing and Social Interaction with Push!Music
- 2007
-
In: Proceedings of the 40th Hawaii International Conference on System Sciences. - Los Alamitos, Calif. : IEEE Computer Society Washington. - 1530-1605. - 0769527558 ; , s. 87-
-
Conference paper (peer-reviewed)abstract
- Push!Music is a novel mobile music listening and sharing system, where users automatically receive songs that have autonomously recommended themselves from nearby players depending on similar listening behaviour and music history. Push!Music also enables users to wirelessly send songs between each other as personal recommendations. We conducted a two-week preliminary user study of Push!Music, where a group of five friends used the application in their everyday life. We learned for example that the shared music in Push!Music became a start for social interaction and that received songs in general were highly appreciated and could be looked upon as 'treats'.
|
|
| 4. |
- Martinsson, John, et al.
(author)
-
Automatic blood glucose prediction with confidence using recurrent neural networks
- 2018
-
In: CEUR Workshop Proceedings. - : CEUR. ; 2148, s. 64-68
-
Conference paper (peer-reviewed)abstract
- Low-cost sensors continuously measuring blood glucose levels in intervals of a few minutes and mobile platforms combined with machine-learning (ML) solutions enable personalized precision health and disease management. ML solutions must be adapted to different sensor technologies, analysis tasks and individuals. This raises the issue of scale for creating such adapted ML solutions. We present an approach for predicting blood glucose levels for diabetics up to one hour into the future. The approach is based on recurrent neural networks trained in an end-to-end fashion, requiring nothing but the glucose level history for the patient. The model outputs the prediction along with an estimate of its certainty, helping users to interpret the predicted levels. The approach needs no feature engineering or data pre-processing, and is computationally inexpensive.
|
|
| 5. |
- Magnusson, Peter S., et al.
(author)
-
SimICS/sun4m : A virtual workstation
- 2019
-
In: USENIX 1998 Annual Technical Conference. - New Orleans, LA, USA : USENIX Association.
-
Conference paper (peer-reviewed)abstract
- System level simulators allow computer architects and system software designers to recreate an accurate and complete replica of the program behavior of a target system, regardless of the availability, existence, or instrumentation support of such a system. Applications include evaluation of architectural design alternatives as well as software engineering tasks such as traditional debugging and performance tuning. We present an implementation of a simulator acting as a virtual workstation fully compatible with the sun4m architecture from Sun Microsystems. Built using the system-level SPARC V8 simulator SimICS, SimICS/sun4m models one or more SPARC V8 processors, supports user-developed modules for data cache and instruction cache simulation and execution profiling of all code, and provides a symbolic and performance debugging environment for operating systems. SimICS/sun4m can boot unmodified operating systems, including Linux 2.0.30 and Solaris 2.6, directly from snapshots of disk partitions. To support essentially arbitrary code, we implemented binary-compatible simulators for several devices, including SCSI, console, interrupt, timers, EEPROM, and Ethernet. The Ethernet simulation hooks into the host and allows the virtual workstation to appear on the local network with full services available (NFS, NIS, rsh, etc). Ethernet and console traffic can be recorded for future playback. The performance of SimICS/sun4m is sufficient to run realistic workloads, such as the database benchmark TPC-D, scaling factor 1/100, or an interactive network application such as Mozilla. The slowdown in relation to native hardware is in the range of 25 to 75 (measured using SPECint95). We also demonstrate some applications, including modeling an 8-processor sun4m version (which does not exist), modeling future memory hierarchies, and debugging an operating system.
|
|
| 6. |
- Michel, Mathieu, et al.
(author)
-
Load-Balanced Data Collection through Opportunistic Routing
- 2015. - 9
-
In: 2015 International Conference on Distributed Computing in Sensor Systems. - 9781479988563 ; , s. 62-70
-
Conference paper (peer-reviewed)abstract
- Wireless Sensor Networks performing low-power data collection often suffer from uneven load distribution among nodes. Nodes close to the network root typically face a higher load, see their battery deplete first, and become prematurely unable to operate (both sensing and relaying other nodes' data). We argue that opportunistic routing, by making forwarding decision on a per-packet basis and at the receiver rather than the sender, has the potential to better balance the load across nodes. We extend ORPL, an opportunistic version of the standard routing protocol RPL, with support for load-balancing. In our protocol, ORPL-LB, nodes continuously adapt their wake-up interval in order to adjust their availability and attain a deployment-specific target duty cycle. We implement our protocol in Contikiand present our experimental validation in Indriya, a 93-nodestestbed. Our results show that ORPL-LB reduces significantly(by approximately 40%) the worst node's duty cycle, with little or no impact on packet delivery ratio and latency.
|
|
| 7. |
- Nazari, N., et al.
(author)
-
Multi-level Binarized LSTM in EEG Classification for Wearable Devices
- 2020
-
In: Proceedings - 2020 28th Euromicro International Conference on Parallel, Distributed and Network-Based Processing, PDP 2020. - : Institute of Electrical and Electronics Engineers Inc.. - 9781728165820 ; , s. 175-181
-
Conference paper (peer-reviewed)abstract
- Long Short-Term Memory (LSTM) is widely used in various sequential applications. Complex LSTMs could be hardly deployed on wearable and resourced-limited devices due to the huge amount of computations and memory requirements. Binary LSTMs are introduced to cope with this problem, however, they lead to significant accuracy loss in some applications such as EEG classification which is essential to be deployed in wearable devices. In this paper, we propose an efficient multi-level binarized LSTM which has significantly reduced computations whereas ensuring an accuracy pretty close to full precision LSTM. By deploying 5-level binarized weights and inputs, our method reduces area and delay of MAC operation about 31× and 27× in 65nm technology, respectively with less than 0.01% accuracy loss. In contrast to many compute-intensive deep-learning approaches, the proposed algorithm is lightweight, and therefore, brings performance efficiency with accurate LSTM-based EEG classification to realtime wearable devices.
|
|
| 8. |
- Täckström, Oscar, et al.
(author)
-
Uncertainty Detection as Approximate Max-Margin Sequence Labelling
- 2010
-
In: CoNLL 2010. - : Association for Computational Linguistics. ; , s. 84-91
-
Conference paper (peer-reviewed)abstract
- This paper reports experiments for the CoNLL 2010 shared task on learning to detect hedges and their scope in natural language text. We have addressed the experimental tasks as supervised linear maximum margin prediction problems. For sentence level hedge detection in the biological domain we use an L1-regularised binary support vector machine, while for sentence level weasel detection in the Wikipedia domain, we use an L2-regularised approach. We model the in-sentence uncertainty cue and scope detection task as an L2-regularised approximate maximum margin sequence labelling problem, using the BIO-encoding. In addition to surface level features, we use a variety of linguistic features based on a functional dependency analysis. A greedy forward selection strategy is used in exploring the large set of potential features. Our official results for Task 1 for the biological domain are 85.2 F1-score, for the Wikipedia set 55.4 F1-score. For Task 2, our official results are 2.1 for the entire task with a score of 62.5 for cue detection. After resolving errors and final bugs, our final results are for Task 1, biological: 86.0, Wikipedia: 58.2; Task 2, scopes: 39.6 and cues: 78.5.
|
|
| 9. |
- Birgersson, Marcus, 1988, et al.
(author)
-
Data Integration Using Machine Learning
- 2016
-
In: Proceedings - IEEE International Enterprise Distributed Object Computing Workshop, EDOCW. - 1541-7719. ; 2016-September, s. 313-322, s. 313-322
-
Conference paper (peer-reviewed)abstract
- Today, enterprise integration and cross-enterprise collaboration is becoming evermore important. The Internet of things, digitization and globalization are pushing continuous growth in the integration market. However, setting up integration systems today is still largely a manual endeavor. Most probably, future integration will need to leverage more automation in order to keep up with demand. This paper presents a first version of a system that uses tools from artificial intelligence and machine learning to ease the integration of information systems, aiming to automate parts of it. Three models are presented and evaluated for precision and recall using data from real, past, integration projects. The results show that it is possible to obtain Fo.5 scores in the order of 80% for models trained on a particular kind of data, and in the order of 60%-70% for less specific models trained on a several kinds of data. Such models would be valuable enablers for integration brokers to keep up with demand, and obtain a competitive advantage. Future work includes fusing the results from the different models, and enabling continuous learning from an operational production system.
|
|
| 10. |
- Laaksolahti, Jarmo, et al.
(author)
-
The lega : A device for leaving and finding tactile traces
- 2011
-
In: Proceedings of the 5th International Conference on Tangible Embedded and Embodied Interaction. - New York, NY, USA : ACM. - 9781450306287 ; , s. 193-196
-
Conference paper (peer-reviewed)abstract
- This paper describes experiences from development and deployment of the Lega, a hand held device for physical sharing of experiences during an art exhibition. Touching and moving the device in different ways creates a tactile trace that can be experienced by others through their own device. The system was successfully deployed at an art exhibition for two months where user studies were performed. Here we present some general observations regarding the systems performance and discuss issues that we encountered.
|
|
