Collaborating Authors

How Testilio wants to rethink software testing


Kruustük: "It struck me that a lot of the testing being done wasn't very good. I could see that there were better ways to do it." Software testing is one of those routine tasks that has to be done to iron out every glitch in a piece of software. For many it's not inspiring stuff, but for Estonian company Testlio, and its CEO Kristel Kruustük, testing has become something of a passion. The company came about after Kruustük was on a team that picked up the $25,000 prize in the 2012 AngelHack, more than enough money to start a new company.

Combinatorial Sleeping Bandits with Fairness Constraints Machine Learning

The multi-armed bandit (MAB) model has been widely adopted for studying many practical optimization problems (network resource allocation, ad placement, crowdsourcing, etc.) with unknown parameters. The goal of the player here is to maximize the cumulative reward in the face of uncertainty. However, the basic MAB model neglects several important factors of the system in many real-world applications, where multiple arms can be simultaneously played and an arm could sometimes be "sleeping". Besides, ensuring fairness is also a key design concern in practice. To that end, we propose a new Combinatorial Sleeping MAB model with Fairness constraints, called CSMAB-F, aiming to address the aforementioned crucial modeling issues. The objective is now to maximize the reward while satisfying the fairness requirement of a minimum selection fraction for each individual arm. To tackle this new problem, we extend an online learning algorithm, UCB, to deal with a critical tradeoff between exploitation and exploration and employ the virtual queue technique to properly handle the fairness constraints. By carefully integrating these two techniques, we develop a new algorithm, called Learning with Fairness Guarantee (LFG), for the CSMAB-F problem. Further, we rigorously prove that not only LFG is feasibility-optimal, but it also has a time-average regret upper bounded by $\frac{N}{2\eta}+\frac{\beta_1\sqrt{mNT\log{T}}+\beta_2 N}{T}$, where N is the total number of arms, m is the maximum number of arms that can be simultaneously played, T is the time horizon, $\beta_1$ and $\beta_2$ are constants, and $\eta$ is a design parameter that we can tune. Finally, we perform extensive simulations to corroborate the effectiveness of the proposed algorithm. Interestingly, the simulation results reveal an important tradeoff between the regret and the speed of convergence to a point satisfying the fairness constraints.

Design in Tech Report 2018


For this year's report, I took a stab at learning all the CSS/JS that I've always wanted to know, and then went after the task of making a fully responsive report. I've succeeded in doing so, and so this PDF version isn't as good as the real thing. In the next few days I will be sharing a link to the real digital experience. But for now -- enjoy this static version of the report which has a few parts that couldn't render to static form. Because ... this year's report is truly computationally designed and therefore needs to be expressed appropriately (smile). Expect a video version on my new YouTube channel "John Maeda is Learning." What can I do about it? As the marginal return on computing power (a la Moore's law) diminishes and technology is less of a differentiating factor, the value of design has entered the foreground. Five (20%) of the top cumulative-funded VC- backed ventures that have raised additional capital since 2013 are noted to have designer co-founders.