# Borealisai.com

## Field Guide to Machine Learning Product Development: …

The design of machine learning requirements takes a careful combination of design and business goals. PMs need to understand the end user to drive value. By focusing on the end-user and developing a deep understanding of a specific problem your product can solve. Otherwise, you run the risk of developing an extremely powerful system only to

Actived: 6 days ago

### Borealis AI Borealisai.com

Heterogeneous Multi-task Learning with Expert Diversity. Oct. 27, 2021. Desired characteristics for real-world RL agents. Oct. 4, 2021. Meet Turing by Borealis AI, an AI-powered text to SQL database interface. June 15, 2021. Tutorial #15: Parsing I: …

Category:  Learn

### Tutorial #3: few-shot learning and meta-learning II

Model agnostic meta-learning or MAML (Finn et al. 2017) is a meta-learning framework that can be applied to any model that is trained with a gradient descent procedure. The aim is to learn a general model that can easily be fine-tuned for many different tasks, even when the training data is scarce.

Category:  Learn

### Designing machine learning for human users

This is part 5 in our Field Guide to Machine Learning Product Development: What Product Managers Need to Know series. Read the Introduction here, learn about how to manage data throughout the machine learning product lifecycle, read up on the Discovery phase here and Feasibility phase here, and stay tuned for more deep dives into the six main stages of the …

Category:  Learn

### Aiden: Reinforcement Learning Applied to

Aiden: Reinforcement Learning Applied to Electronic Trading. RBC Capital Markets and Borealis AI have developed an AI-powered electronic trading platform with the goal of delivering improved execution quality and insights for our clients globally. Aiden uses Deep Reinforcement Learning to learn from its experiences in the market and adjust to

Category:  Learn

### Tutorial #17: Transformers III: Training

Learning rate warm-up (in which the learning rate is gradually increased during the early stages of training) is particularly puzzling. This is not required for most deep learning architectures. However, training fails for transformers if we just start with a typical learning rate.

Category:  Learn

### Machine Learning Researcher

Borealis AI, a RBC Institute for Research, is a curiosity-driven research centre dedicated to achieving state-of-the-art in machine learning. Established in 2016, and with labs in Toronto, Montreal, Waterloo and Vancouver, we support academic collaborations and partner with world-class research centres in artificial intelligence.

Category:  Learn

### Tutorial #10: SAT Solvers II: Algorithms

Other examples of machine learning in the SAT community include Nejati et al., (2017) where reinforcement learning is used to decide when to restart the solver. Discussion. In this blog, we introduced resolution and conditioning operations. We then developed a series of algorithms based on these operations.

Category:  Learn

### Tutorial #9: SAT Solvers I: Introduction and applications

Second, machine learning techniques are often used as components of SAT solvers; in part II of this tutorial, we'll discuss how reinforcement learning can be used to speed up SAT solving, and in part III we will show that there is a close connection between factor graphs and SAT solvers and that belief propagation algorithms can be used to

Category:  Learn

### Tutorial #13: Differential privacy II: machine learning

Tutorial #13: Differential privacy II: machine learning and data generation. March 19, 2021. Authors: G. Sharma , N. Hegde, S. Prince , M. Brubaker. In part I of this tutorial we discussed what it means for data analysis to be "private" with respect to a dataset and argued that perfect privacy is not possible.

Category:  Learn

### How to navigate the model and product development phase of

Navigating the uncertainties of the machine learning model development phase takes a PM skill set that goes beyond fluency with scrum practices, clarity of vision, and clear communication. It requires techniques for creating a sense of team unity and momentum (even when it's unclear when and how things will work).

Category:  Learn

As a Machine Learning Engineering Lead, you’ll be responsible for owning and delivering a project end to end – everything from data pre-processing to implementing machine learning training pipelines and deploying inference in a production environment. You’ll be directly managing a team of machine learning engineers, responsible for

Category:  Learn

### Using Config Files to run Machine Learning Experiments

Here are some guidelines to do so successfully: Identify all mutable parameters in your machine learning project and write them in the config file. Consider writing the model parameters, the loss function, the dataset, the data loader, the metric, the optimizer, and the learning rate scheduler. The config file will help you quickly explore

Category:  Learn

### On building energy efficient networks

Machine learning models are helping to solve increasingly difficult tasks, but as a natural result of scale, some of the models are getting enormous. Often, our community creates models that work just for the particular task at hand; but if these techniques are to be widely deployable, we must work to decrease the energy of these models.

Category:  Learn

### Model Validation: a vital tool for building trust in AI

It is widely known that machine learning algorithms tend to perform worse on images of women and people of colour [1]. To some extent this is due to biased datasets; however, we must be vigilant at all stages of the ML lifecycle. We need to rethink how we design, test, deploy, and monitor machine learning algorithms.

Category:  Learn

### Aiden – Reinforcement learning for order execution

Aiden reinforcement learning overview. Aiden predicts the policy which is a distribution over several possible actions that relate to the aggressiveness of the order. One action is drawn randomly from this distribution and this is then converted to a limit order.

Category:  Learn

### Tutorial #4: auxiliary tasks in deep reinforcement learning

Reinforcement learning (RL) can now produce super-human performance on a variety of tasks, including board games such as chess and go, video games, and multi-player games such as poker. However, current algorithms require enormous quantities of data to learn these tasks. For example, OpenAI Five generates 180 years of gameplay data per day, and

Category:  Learn

### The Borealis AI 2020-2021 Fellowships: Supporting Canada’s

The Borealis AI 2021 Fellowships have been awarded to: “These Borealis AI Fellowships are a strong endorsement of the hard work being done at Canada’s Universities and Machine Learning Research Institutes. More importantly, they directly support Canadian research and research teams – like those at Dalhousie University – as they strive

Category:  Learn

### Learning from Young Pioneers: A Look inside the AI4Good

They reminded us of the importance of social good and life-long learning. For many of the government and industry partners in attendance at the AI4Good Lab’s Industry Night organised by CIFAR this year, the plan was to provide support, mentorship and guidance to the participants of the AI4Good Lab – a Canadian AI training initiative for

Category:  Learn

### AI as a Personal Banker

The machine learning challenge. We’re exploring a few key areas of machine learning research to apply prediction solutions to personal banking. Most banking data has a timestamp attached to it, denoting when an event such as a purchase took place. But those times are often irregular.

Category:  Learn

### Improving reinforcement learning with human input

Reinforcement learning (RL) is a popular type of machine learning that allows an agent to interact with an environment. Rather than being told if its actions are correct or incorrect, it is only provided a (possibly time delayed and stochastic) reward signal that it seeks to maximize.The RL framework allows agents to learn how to play video games, control robots, and optimize data …

Category:  Learn

### RESPECT AI: Resetting regulation: A new approach to

In this post, we hear Gillian Hadfield’s views on the need for new regulation and new regulatory approaches for machine learning. Gillian is the director of the Schwartz Reisman Institute for Technology and Society; professor of law and of strategic management at the University of Toronto; a faculty affiliate at the Vector Institute for Artificial Intelligence; and a …

Category:  Learn

### Summer 2022 Research Internships

With 40+ scientific publications in top-tier academic venues, the institute performs research in areas, such as deep learning, reinforcement learning, language processing, AI safety, and more. Borealis AI was founded in 2016 and has over 100 team-members across its four labs in Canada. Our internships are currently running virtually.

Category:  Learn

### Tutorial #12: Differential Privacy I: Introduction

The recent machine learning revolution has been fueled by data, and this data often comes from the end-users themselves. Consider a smartphone manufacturer trying to make typing faster and more accurate, a store trying to help its customers find the products they want, or a doctor trying to interpret a medical test.

Category:  Learn

### Applying differential privacy to deep reinforcement learning

While deep learning applications have successfully integrated into multiple product categories, RL has had a slower initiation. The recent momentum behind RL-based commercialization has been propelled by research advancements that have naturally lent themselves to product ideas in specific sectors, like financial markets, health care and marketing.

Category:  Learn

### Intelligibility is a key component to trust in machine

This post argues that intelligibility is a key component to trust. 1 The deep learning explosion has brought us many high-performing algorithms that can tackle complex tasks at superhuman levels (e.g., playing the games of Go and Dota 2, or optimizing data centers ). However, a common complaint is that such methods are inscrutable “black

Category:  Learn

### Tutorial #5: variational autoencoders

During learning we are given training data $\{\mathbf{x}_{i}\}_{i=1}^{I}$ and want to maximize the log likelihood of the model with respect to the parameters $\boldsymbol\phi$. For simplicity we'll assume that the variance term $\sigma^2$ in the likelihood expression is known and just concentrate on learning $\boldsymbol\phi$: \begin{eqnarray}

Category:  Learn

### Tutorial #7: neural natural language generation

Reinforcement learning setup. Let's briefly recap the reinforcement learning framework. In RL an agent performs actions in an environment and observes the consequences of these actions through (i) changes in the environment and (ii) a numerical reward which indicates whether the agent is achieving its task or not.

Category:  Learn

### Is multiagent deep reinforcement learning the answer or

Deep reinforcement learning (DRL) is a recent yet very active area of research that joins forces between deep learning (the use of neural networks) and reinforcement learning (solving sequential decision tasks). In DRL, the goal is to learn an optimal policy (behavior) of an agent acting in an environment, with deep neural networks as the

Category:  Learn

### Machine Learning Researcher

Borealis AI, a RBC Institute for Research, is a curiosity-driven research centre dedicated to achieving state-of-the-art in machine learning. Established in 2016, and with labs in Toronto, Montreal, Waterloo and Vancouver, we support academic collaborations and partner with world-class research centres in artificial intelligence.

Category:  Learn

### ICML 2020 Roundup

ICML 2020 Roundup. In July the Thirty-seventh International Conference on Machine Learning (ICML) was held featuring over 1,000 papers and an array of tutorials, workshops, invited talks and more. Borealis researchers were (virtually) there presenting their work. and many members of the research team took the time to virtually attend ICML 2020.

Category:  Learn

### Improving Reinforcement Learning with Human Input

3.3 Shaping with Inverse Reinforcement Learning Rather than using collected demonstrations to learn a clas-siﬁer and then change the action-selection method, consider instead how the actual reward could be changed. In particu-lar, Inverse Reinforcement Learning (IRL) is the problem of learning a reward function using a set of observations from

Category:  Learn

### Machine Learning Researcher

As a Machine Learning Researcher in our capital markets team, you’re looking to channel your love of playing with real-world data into industry-disrupting solutions. Our team supports research on a wide variety of theoretical and applied machine learning projects, including game-changing studies on dynamic interactions between statistical

Category:  Learn

### Machine Learning Reading Group: The Mechanics of n-Player

Machine Learning Reading Group: The Mechanics of n-Player Differentiable Games. One of the driving factors behind the modern deep learning explosion can be attributed to the success of training neural networks with stochastic gradient descent (SGD). We form a neural network as a composition of functions, apply an auto-differentiation library

Category:  Learn

### Machine Learning Researcher

Proficiency in Python and Deep Learning packages such as Tensorflow or PyTorch. Demonstrated ability to deliver machine learning solutions following industry best practices in software engineering is an asset. Experience with time series or event forecasting, especially with asynchronous data, is an asset.

Category:  Learn

### Our key takeaways from ICLR 2018

Machine Learning in the Real World. The invited talk by Suchi Saria (“Augmenting Clinical Intelligence with Machine Intelligence“) gave an excellent example of the difficulties in applying machine learning algorithms on applications where supervised data is unavailable or highly skewed. In this case, Prof. Saria looked at patient survival

Category:  Learn

### Research Engineer – Data Systems

Research Engineer Data Systems. Borealis AI is a team of researchers and developers dedicated to solving today’s leading problems in machine learning and artificial intelligence. Our researchers are dedicated to pushing the boundaries of theoretical and applied science, while our development team transforms state-of-the-art technologies

Category:  Learn

### Research Tutorials and Recommended Resources for the Avid

Tutorial #4: auxiliary tasks in deep reinforcement learning. This tutorial focuses on the use of auxiliary tasks to improve the speed of learning in the context of deep reinforcement learning (RL). Auxiliary tasks are additional tasks that are learned simultaneously with the main RL goal and that generate a more consistent learning signal.

Category:  Learn

### Robust Risk-Sensitive Reinforcement Learning Agents for

Then, we augment the framework to multi-agent learning and assume an adversary which can take over and perturb the learning process. Our third and fourth algorithms perform well under this setting and balance theoretical guarantees with practical use. Additionally, we consider the multi-agent nature of the environment and our work is the first

Category:  Learn

### Our key takeaways from ICML 2017

Meta-learning is not meant to replace systems like unsupervised learning, or semi-supervised learning; rather, it provides a new perspective and set of techniques to help solve these existing problems. Today, when you’re given data, you have one model that learns the structure of the data, while higher-level models inform us of the best and

Category:  Learn