4,752 research outputs found
Biologically Plausible Reinforcement Learning
No full textRoelfsema, P.R. [Promotor]Bohte, S.M. [Copromotor
Competitive Market-based Allocation of Consumer Attention
No full textThe amount of attention space available for recommending suppliers to consumers on e-commerce sites is typically limited. We present a competitive distributed recommendation mechanism based on adaptive software agents for efficiently allocating the "consumer attention space", or banners. In our approach, each agent bids in an auction for the momentary attention of each consumer. Successive auctions allow agents to rapidly adapt their bidding strategy to focus on consumers interested in their offerings. We demonstrate the feasibility of our system by an evolutionary simulation, and reflect on the advantages of this distributed market-based approach
Market-based Recommendation: Agents that Compete for Consumer Attention
No full textThe amount of attention space available for recommending suppliers to consumers on e-commerce sites is typically limited. We present a competitive distributed recommendation mechanism based on adaptive software agents for efficiently allocating the 'consumer attention space', or banners. In the example of an electronic shopping mall, the task is delegated to the individual shops, each of which evaluates the information that is available about the consumer and his or her interests (e.g. keywords, product queries, and available parts of a profile). Shops make a monetary bid in an auction where a limited amount of 'consumer attention space' for the arriving consumer is sold. Each shop is represented by a software agent that bids for each consumer. This allows shops to rapidly adapt their bidding strategy to focus on consumers interested in their offerings. For various basic and simple models for on-line consumers, shops, and profiles, we demonstrate the feasibility of our system by evolutionary simulations as in the field of agent-based computational economics (ACE). We also develop adaptive software agents that learn bidding strategies, based on neural networks and strategy exploration heuristics. Furthermore, we address the commercial and technological advantages of this distributed market-based approach. The mechanism we describe is not limited to the example of the electronic shopping mall, but can easily be extended to other domains
Automated Negotiation and Bundling of Information Goods
No full textIn this paper, we present a novel system for selling bundles of news items. Through the system, customers bargain with the seller over the price and quality of the delivered goods. The advantage of the developed system is that it allows for a high degree of flexibility in the price, quality, and content of the offered bundles. The price, quality, and content of the delivered goods may, for example, differ based on daily dynamics and personal interest of customers. Autonomous "software agents" execute the negotiation on behalf of the users of the system. To perform the actual negotiation these agents make use of bargaining strategies. We present the novel approach of decomposing bargaining strategies into concession strategies and Pareto-efficient search strategies. Additionally, we introduce the orthogonal and orthogonal-DF strategy: two Pareto search strategies. We show through computer experiments that the use of these Pareto search strategies will result in very efficient bargaining outcomes. Moreover, the system is setup such that it is actually in the best interest of the customer to have their agent adhere to this approach of disentangling the bargaining strategy
Predicting periodic and chaotic signals using Wavenets
No full textThis thesis discusses forecasting periodic time series using Wavenets with an application in financial time series. Conventional neural networks used for forecasting such as the LSTM and the full convolutional network (FCN) are computationally expensive. The Wavenet uses dilated convolutions which significantly reduces the computational cost compared to the FCN with the same number of inputs. Forecasts made on the sine wave shows that the network can successfully fully forecast a sine wave. Forecasts made on the Mackey Glass time series shows that the network can outperform the LSTM and other methods Furthermore, forecasts made on the Lorenz system shows that the network is able to outperform the LSTM. By conditioning the network on the other relevant coordinate, the prediction becomes more accurate and is able to make full forecasts. In a financial application, the network shows less predictive accuracy compared to multivariate dynamic kernel support vector machines. <br/
Efficient Spike-Coding with Multiplicative Adaptation in a Spike Response Model
No full textNeural adaptation underlies the ability of neurons to maximize encoded informa- tion over a wide dynamic range of input stimuli. While adaptation is an intrinsic feature of neuronal models like the Hodgkin-Huxley model, the challenge is to in- tegrate adaptation in models of neural computation. Recent computational models like the Adaptive Spike Response Model implement adaptation as spike-based addition of fixed-size fast spike-triggered threshold dynamics and slow spike- triggered currents. Such adaptation has been shown to accurately model neural spiking behavior over a limited dynamic range. Taking a cue from kinetic models of adaptation, we propose a multiplicative Adaptive Spike Response Model where the spike-triggered adaptation dynamics are scaled multiplicatively by the adap- tation state at the time of spiking. We show that unlike the additive adaptation model, the firing rate in the multiplicative adaptation model saturates to a maxi- mum spike-rate. When simulating variance switching experiments, the model also quantitatively fits the experimental data over a wide dynamic range. Furthermore, dynamic threshold models of adaptation suggest a straightforward interpretation of neural activity in terms of dynamic signal encoding with shifted and weighted exponential kernels. We show that when thus encoding rectified filtered stimulus signals, the multiplicative Adaptive Spike Response Model achieves a high coding efficiency and maintains this efficiency over changes in the dynamic signal range of several orders of magnitude, without changing model parameters
Hersenen simuleren voor computer
No full textOns brein kent 100 miljard neuronen en dat netwerk verslaat ‘gewone computers’. Simulaties van de hersenen kunnen de IT dus veel leren. Het CWI slaagde erin complexe hersenactiviteit te simuleren op eenvoudige grafische kaarten. Dat versnelt onderzoek en ontwikkeling van die namaak-breinen
- …
