About
Activity
594 followers
-
Benjamin Bossan liked thisThis post was an absolute blast! If you are writing #pytorch training/validation loops you should take a look at those libraries and see how much time you can save. I hope you will enjoy this as much as I have.Benjamin Bossan liked thisWe have something pretty awesome for all the #pytorch users out there! We got creators and core contributors behind ALL of the model training libraries in PyTorch Ecosystem to talk about their projects, specifically: - Philosophy - API structure - The learning curve for new users - Build-in features (what you get out-of-the-box) - Extension capabilities (simplicity of integration in research) - Reproducibility - Distributed training - Productionalization - Popularity The result was this blog post: http://bit.ly/389KfxL Hopefully, with this article, it will be easier to choose some of those libraries to improve your #deeplearning projects! If you have any additional questions, stuff we didn’t mention that is interesting to you, drop a comment! Huge thanks to: - William Falcon from PyTorch Lightning - Sergey Kolesnikov and the Catalyst Team - Benjamin Bossan from Skorch - Victor FOMIN from #pytorch Ignite - Sylvain Gugger and Jeremy Howard from fast.ai - Ethan Harris and Matt Painter from Torchbearer8 Creators and Core Contributors Talk About Their Model Training Libraries From PyTorch Ecosystem8 Creators and Core Contributors Talk About Their Model Training Libraries From PyTorch Ecosystem
-
Benjamin Bossan liked thisIt was so much fun creating this integration. Thank you so much Benjamin Bossan Thomas J Fan!Benjamin Bossan liked thisNeptune is now integrated with Skorch (scikit-learn compatible #pytorch wrapper)! With just 3 lines of code: experiment = neptune.create_experiment(...) logger = NeptuneLogger(exp) net = NeuralNetClassifier(..., callbacks=[logger]) you log: - metrics/losses - code - resource consumption - hyperparameters And you can log more with an additional line or two! Check out the docs: http://bit.ly/2SX8itS #MachineLearning #deeplearning #datascience
-
Benjamin Bossan liked thisThanks Rui Mendes Leonid K. David Campos and the rest of the DSPT Day 2019 organization team for such a great event. Great organization, awesome people and a lot knowledge exchange. Thanks for allowing me to be part of this.Benjamin Bossan liked thisMiguel Cabrera talked about Data Science in a traditional fashion retailer + Exploring Demand Forecasting. Demand forecasting is key for the retailing industry. This is especially true in the fashion industry where product demand is volatile and the life cycle is short. Miguel described the Demand Forecasting Problem in the context of a brick-and-mortar retailer and discussed the commonly used techniques, including evaluation metrics, feature and target transformations and commonly used predictors. There was also time for a short discussion on the challenges of successfully implementing a demand forecasting solution beyond the technical details. #DSPTDay #DSPTDay2019 #bigdata #data #machinelearning #analytics #datascienceportugal #conference #porto #dspt #artificialintelligence
-
Benjamin Bossan liked thisBenjamin Bossan liked thisToday Marian one of our data scientists is presenting his development of #skorch at the #pytorch dev conference. Thanks Benjamin Bossan and @Marian for your development. We love #skorch here Otto Group data.works. https://lnkd.in/dN6wdMt
Experience & Education
Licenses & Certifications
Publications
-
The evolution of social learning and its economic consequences
Journal of Economic Behavior & Organization
We use an evolutionary model to simulate agents who choose between two options with stochastically varying payoffs. Two types of agents are considered: individual learners, who rely on trial-and-error methods, and social learners, who imitate the wealthiest sampled individual. Agents adapt to changing environments within one generation by using their respective learning strategy. The frequency of the agent types adapts between generations according to the agents’ acquired wealth. During the…
We use an evolutionary model to simulate agents who choose between two options with stochastically varying payoffs. Two types of agents are considered: individual learners, who rely on trial-and-error methods, and social learners, who imitate the wealthiest sampled individual. Agents adapt to changing environments within one generation by using their respective learning strategy. The frequency of the agent types adapts between generations according to the agents’ acquired wealth. During the course of evolution, social learning becomes dominant, resulting in three major effects: First, for better or worse, the decisions of social learners are more exaggerated than those of individual learners. Second, social learners react with a delay to changes in the environment. Third, the behavior of social learners becomes more and more detached from reality. We argue that our model gives insights into economic systems and markets.
Other authors -
We were all young once: an intragenomic perspective on parent–offspring conflict
Proceedings of the Royal Society B: Biological Sciences
-
Evolutionary Perspectives on Social Cognition
Strüngmann Forum Reports
Honors & Awards
-
2nd place of 2225: Springleaf $100.000 competition
Kaggle
https://www.kaggle.com/c/springleaf-marketing-response/leaderboard
Languages
-
German
-
-
English
-
-
French
-
-
Latin
-
Other similar profiles
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content