Reinforcement learning’s (RL) objective is to develop agents able to learn optimal policies in unknown environments by trial-and-error and with limited supervision. Recent developments in exploration-exploitation, online learning, and representation learning are making RL more and more appealing to real-world applications, with promising results in challenging domains such as recommendation systems, computer games, and robotics. The 12th edition of EWRL will serve as a forum to discuss the state-of-the-art and future research directions and opportunities for the growing field of RL. Beyond traditional topics, we will encourage discussions on representation learning, risk-averse learning, apprenticeship and transfer learning, and practical applications.
https://ewrl.wordpress.com/ewrl12-2015/

Deep learning is a fast-growing field of Machine Learning concerned with the study and design of computer algorithms for learning good representations of data, at multiple levels of abstraction. There has been rapid progress in this area in recent years, both in terms of methods and in terms of applications, which are attracting the major IT companies. Many challenges remain, however, in aspects like large-scale (hyper-) parameter optimization, modeling of temporal data with long-term dependencies, generative modeling, efficient Bayesian inference for deep learning, multi-modal data and models, and learning representations for reinforcement learning. The workshop aims at bringing together researchers in the field of deep learning to discuss recent advances, ongoing developments and the road that lies ahead.
https://sites.google.com/site/deeplearning2015/

Active learning has been a topic of significant research over the past several decades with much attention devoted to both theoretical and practical considerations. A variety of algorithms and sampling paradigms have been proposed and studied, but roughly speaking, this line of research focuses on how to make feedback driven decisions about data collection, and how to leverage this power for efficient learning. In this workshop, we hope to find future research directions that address the disconnect between active learning theory and practice.
https://sites.google.com/site/icmlalworkshop/

Constructive machine learning describes a class of machine learning problems where the ultimate goal is not finding a good model of the data but rather one or more particular instances of the domain which are likely to exhibit desired properties. While traditional approaches choose these instances from a given set of unlabeled instances, constructive machine learning is typically iterative and searches an infinite or exponentially large instance space. With this workshop we want to bring together domain experts employing machine learning tools in constructive processes and machine learners investigating novel approaches or theories concerning constructive processes as a whole.
http://www.kdml-bonn.de/cml2015

Crowdsourcing and human computation are emerging paradigms in computing impacting the ability of academic researchers to build new systems and run new experiments involving people, and is also gaining a lot of use within industry for collecting training data for the purpose of machine learning. The fundamental question that we plan to explore in this workshop is: How can we build systems that combine the intelligence of humans and the computing power of machines for solving challenging scientific and engineering problems? The goal is to improve the performance of complex human-powered systems by making them more efficient, robust, and scalable.
http://crowdwisdom.cc/icml2015/

There is an increasing number of real world classification tasks (e.g. web, biology) where the number of labels is very large (thousands or even millions), the labels are noisy, many of them are rare, sub-linear time prediction (in the number of labels) is mandatory, among several other challenges. This problem, called Extreme Classification is the central topic of the workshop: How to solve it efficiently ?
https://sites.google.com/site/extremeclassification/

Data consisting of an underlying discrete structure associated with continuous attributes is becoming increasingly important. Representing objects by sets, graphs or sequences is relevant for many fields such as natural language processing, medical imaging, computer vision, bioinformatics, or network analysis. In spite of recent efforts on classification and mining of structural data, many open problems remain. These include converting raw data to a structural representation, combining structure with continuous-valued attributes for classification and regression tasks, and the low sample size / high dimensionality situations associated with biomedical applications, among others. FEAST aims to spark discussion and interaction around these problems.
https://sites.google.com/site/feast2015/

Many problems from machine learning and signal processing have aim at automatically learning sparse representations from data. Aiming at sparsity actually involves an l_0 “norm” regularization/constraint, and the l_1 convex relaxation way is essentially a proxy to induce sparsity. Greedy methods constitute a strategy to tackle the combinatorial optimization problems posed by the issue of learning sparse representations/predictors. This family of methods has been much less investigated than the convex relaxation approach by the ICML community. This is precisely the purpose of this workshop to give a central place to greedy methods for machine learning and discuss the blessings of such methods.
https://sites.google.com/site/gretaproject/greed-is-great-icml

In the last decade, machine learning has had a growing influence on neuroimaging data handling and analysis, making it an ubiquitous component of all kinds of data analysis procedures and software. While it is clear that machine learning has the potential to revolutionize both scientific discovery and clinical diagnosis applications, continued progress requires close collaboration between statisticians, machine learning practitioners and neuroscientists. We propose a one day workshop to provide a forum for interaction between these groups. Our workshop goals are to highlight best practices, disseminate the state of the art in high dimensional methods and related tools with a focus on application to neuroimaging data analysis, and to facilitate discussions to identify the key open problems and opportunities for machine learning in neuroscience.
https://sites.google.com/site/stamlins2015/home

The goal of this workshop is to bring together experts from different fields of machine learning, cognitive science, and education to explore the interdisciplinary nature of research on the topic of machine learning for education. In particular, we aim to elicit new connections among these diverse fields, identify novel tools and models that can be transferred from one to the others, and explore novel machine learning applications that will benefit the education community. Topics of interest of this workshop include learning and content analytics, scheduling, automatic grading systems, cognitive psychology, and experimental design.
http://dsp.rice.edu/ML4Ed_ICML2015

The workshop is about open source software (OSS) in machine learning. Our aim is to bring together developers to share experiences in publishing ML methods as OSS and to foster interoperability between different packages, as well as allowing developers to demonstrate their software to potential users in the machine learning community. Continuing the tradition of previous MLOSS workshops, we will have a mix of invited speakers, contributed talks and discussion/activity sessions. For 2015, we focus on building open ecosystems.
https://mloss.org/workshop/icml15/

The success of machine learning in many domains crucially relies on human machine learning experts, who select appropriate features, workflows, machine learning paradigms, algorithms, and their hyperparameters. The rapid growth of machine learning applications has created a demand for off-the-shelf machine learning methods that can be used easily and without expert knowledge. We call the resulting research area that targets progressive automation of machine learning AutoML. For example, a recent instantiation of AutoML we’ll discuss is the ongoing ChaLearn AutoML challenge (http://codalab.org/AutoML).
http://icml2015.automl.org

Demand forecasting is the problem of predicting the amount of goods or services demanded by customers during some future time range: a critical application for many businesses. Retailers base in-stock management decisions like ordering and storage, as well as supply chain management, on demand forecasts. Energy utility companies use forecasting for scheduling operations, investment planning and price bidding. The data revolution creates new opportunities to improve forecast accuracy and granularity, given that heterogenous data sources can be integrated. The focus of the workshop is on demand forecasting by means of data-driven techniques, with a specific emphasis on retail, energy, and transportation industries. We hope to identify the most important challenges from a business point of view, and to start a focussed discussion on how to formalize and solve them by means of machine learning techniques, or on which tools are missing and require additional research efforts.
https://sites.google.com/site/icmldemand/

This interdisciplinary workshop will consider the issues of fairness, accountability, and transparency in machine learning. It will address growing anxieties about the role that automated machine learning systems play in consequential decision-making in such areas as commerce, employment, healthcare, education, and policing.
http://fatml.org

Kernel methods, such as SVMs and Gaussian processes, provide a flexible and expressive learning framework – they are widely applied to small to medium-size datasets, but have been perceived as lacking in scalability. Recently, there has been a resurgence of interest in various fast approximation techniques for scaling up those methods. Indeed, characterizing tradeoffs between their statistical and computational efficiency is growing into an active research topic with encouraging results: it has been reported that appropriately scaled up kernel methods are competitive with deep neural networks. The workshop will overview recent advances and discuss opportunities and challenges within the field.
https://sites.google.com/site/largescalekernelwsicml15/

The ever-increasing size and accessibility of vast music libraries has created a demand more than ever for machine learning systems that are capable of understanding and organizing this complex data. Collaborative filtering provides excellent music recommendations when the necessary user data is available, but these approaches also suffer heavily from the cold-start problem. Furthermore, defining musical similarity directly is extremely challenging as myriad features play some role (e.g., cultural, emotional, timbral, rhythmic). The topics discussed will span a variety of music recommender systems challenges including cross-cultural recommendation, content-based audio processing and representation learning, automatic music tagging, and evaluation.
http://sites.google.com/site/ml4md2015/

We aim to present original methods and applications on the interface between Machine Learning and Medical Imaging. Developments in machine learning have created novel opportunities in knowledge discovery, analysis, visualisation and reconstruction of medical image datasets. However, medical images also pose several particular challenges for standard approaches, for instance, lack of data availability, poor image quality or dedicated training requirements, giving rise to questions such as how to better exploit smaller datasets, or understand fundamentals on image spaces or generative models. The workshop will cover both theoretical aspects as well as effective applications of machine learning and medical imaging.
https://sites.google.com/site/icml2015mi/

We are gradually moving towards a smart city era. Technologies that apply machine learning algorithms to urban data will have significant impact in a lot of aspects of the citizens’ everyday life. Unfortunately, urban data have some characteristics that hinder the state of the art in machine learning algorithms, such as diversity, privacy, lack of labels, noise, complementarity of multiple sources and requirement for online learning. Many smart city applications require to tackle all these problems at once. This workshop aims at discussing a set of new Machine Learning applications and paradigms emerging from the smart city environment.
http://insight-ict.eu/mud2/

The aim of this workshop is to bring together researchers in the emerging topics related to learning and decision ­making under budget constraints in uncertain and dynamic environments. These problems introduce a new trade off between prediction accuracy and prediction cost. Studying this tradeoff is an inherent challenge that needs to be investigated in a principled fashion in order to invent practically relevant machine learning algorithms.
https://sites.google.com/site/icml2015budgetedml/

Learning systems or robots that interact with their environment by perceiving, acting and communicating often face a challenge in how to bring these different concepts together. The challenge arises because core concepts are still predominantly studied in their core communities, such as the computer vision, robotics or natural language processing communities, without much interdisciplinary exchange. Machine learning lies at the core of these communities, and can therefore act as a unifying factor in bringing them closer together. This will be highly important for understanding how state-of-the-art approaches from different disciplines can be combined, refined, and applied to form generally intelligent interactive systems. It will also open a channel for communication and collaboration across research communities.
http://mlis-workshop.org/2015