Jean-François Djoufak
Senior Manager | Tax

Canada, and specifically, Montréal, buzzes with massive investments and numerous initiatives in artificial intelligence (AI). This is why Raymond Chabot Grant Thornton started AI and advanced analytics practice this year to help SMEs take advantage of this latest technological revolution.

I’ll be leading the section on AI in the Tax-R&D practice’s newsletter. I’ll be delving into the questions you should be asking, practices to adopt, examples that could serve as inspirations for your business. I’ll also be looking at disruptive technologies (ones that displace established technologies and shake up the industry or ground-breaking products that create a completely new industry) that could improve, enhance or even revolutionize business practices.

Marvin Lee Minsky, one of the creators of AI, defined it as the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition and decision-making. In this first article, we’ll be looking at using AI and machine learning to create agents (autonomous software) able to think and reason like humans.

What is machine learning?

Machine learning is a set of techniques (linear regression, Bayesian classification, boosting, neural networks, etc.) used to give a machine the ability to learn from past experiences so it can deduce rules that will form new knowledge and serve as a basis to analyze new situations. For example, using analytical data collected by ecommerce websites, a machine learning algorithm can determine the rules that characterize users most likely to delete their account. Using those rules, the algorithm can analyze users’ actions to offer promotions just before they would take the critical decision. If the calculations of former rules are systematically based on the matchings, set by experts, between analytical data and users, the learning process is said to be supervised. Semi-supervised learning occurs when rules’ calculations partially rely on matchings set by experts. When there are no matchings used in the calculations, the learning process is unsupervised. In the case of reinforced learning, the algorithm results are reused to guide the calculation of the next predictions.

A neural network is a network of computing units (neurons) operating in parallel and arranged in tiers to perform complex functions. Each successive tier uses the output received from the tier preceding it. When a neural network is used for visual recognition, the first tiers identify, for example, lines, curves and angles, the middle tiers identify shapes while the last tiers identify objects such as eyes or wheels. The resulting automatic learning method is known as deep learning. Although deep learning has existed from some 10 years, it has grown in leaps and bounds with the increase in computational power, in particular, the ability to use GPU-accelerated computing (graphics processing unit) for general processing and the advent of large databases.

State a machine learning problem

Machine learning can be used in business to resolve several problems. In order to state the problem properly, you need to determine to which of the four categories of machine learning algorithms it fits.

Raymond Chabot Grant Thornton - image

Classification: The goal of problems in this category is to classify or label each object in a data set in supervised or semi-supervised mode, i.e. with the help of experts. They can be unary (e.g. can unusual client transactions be detected?), binary (e.g. will lead Doe be converted or not?) or multiple (e.g. what type of product is user John most likely to buy: computer, portable or smartphone?).

Clustering: The main difference with problems in this category is that there is no human intervention for determining the classification rules and classes. In order words, it is an unsupervised learning process. Problems in this category include portioning of users for marketing purposes (e.g., what are our main market segments, based on our clients’ demographics and their purchases?) or understanding their behaviour (e.g., how can we classify the key words used for searches in our website?).

Regression: This category includes problems to predict or calculate numeric values rather than classes. This includes price calculations (e.g. what is a fair sale price considering the various production constraints?), product demand predictions (e.g., considering the last marketing campaign, how many products will we sell next month?), etc.

Ranking: In this category, the importance of an object compared with other objects in the same data set is calculated. Examples include recommendations (what five products should be displayed for a user, based on purchasing history?), website layout (how should displays be organized on the website considering users’ browsing history?), etc.

So, the answer to the question in the title is YES. Artificial intelligence can be embedded into your business as long as you can state your problems in terms of the four categories. Initially, you should start with simple projects with considerable added value for your business. Stating the problem is all the more important because it will influence the data collection procedures you will need to implement, which we’ll be seeing in the forthcoming article, with emphasis on the quality and relevance of captured data.

01 Nov 2017  |  Written by :

Jean-François Djoufak is a tax expert at Raymond Chabot Grant Thornton. Contact him today.

See the profile

Next article

Julie Ruel
Manager | Tax

Human beings have been suffering from allergies since the dawn of time. While they weren’t very common for centuries, there has been a significant increase in allergies in the past few decades. Recent discoveries, however, could change things and pave the way for therapeutic remedies.

Allergies on the rise

Over a quarter million Quebecers suffer from food allergies, that is, 4% of the adult population and 8% of the juvenile population. Alarming fact: this number tends to double every five years.

Allergies are a spontaneous, unusual reaction of the immune system when it comes into direct contact with a pathogen. For reasons that are not fully known, the body has an exaggerated response and produces specific antibodies that cause allergy symptoms, which can sometimes be fatal.

Many are questioning this significant rise in allergies. While they are numerous, the causes of this kind of hypersensitivity are not clearly defined. Recent studies lead us to believe that pollution and changes in the food processing industry could be responsible. Access to prepared meals containing a multitude of processed foods may increase the contact with allergens.

Furthermore, these changes could trigger the mutation of natural molecules in food allergens. What’s more, the combination of these different foods might directly affect the allergic intensity coefficient. According to a study conducted in 2003 by English researchers, combining soy milk with peanuts increases the risk of a peanut allergy by 2.6.

The growing increase could also be explained by any of the following: a massive consumption of antibiotics, a decrease in breast feeding and sterile environments. While microbes normally act as indicators for tolerating foreign agents in the body, their decreased contact with the immune system may cause our immune system get out of sync and fight against usually harmless agents such as food.

Effective treatments on the horizon

Other than avoiding allergenic foods, the EpiPen is the most popular emergency treatment for severe allergic reactions. However, the relief is temporary and researchers are seeking to better understand the allergenic factors involved and develop more permanent treatments, which may not totally eliminate allergy symptoms, but could increase the body’s tolerance.

The year 2017 was prolific in terms of allergy-related scientific discoveries. In the spring, researcher Dr. Lamia L’Hocine, from the Saint-Hyacinthe Research and Development Centre, Agriculture and Agri-Food Canada, determined that boiled peanuts contain proteins that break down very quickly in the digestive system compared to roasted peanuts, which have a greater resistance to digestion. Surprisingly, roasted peanuts have shown increased resistance to digestion when mixed with fatty foods and sugar (example: cookie dough).

Despite the fact that these results were obtained in vitro, the findings tend to indicate that boiled peanuts might limit allergic reactions. In fact, allergic reactions go through two distinct, but often corollary phases: direct contact and sensitization of the gastro-intestinal immune system. It could therefore be possible that consuming boiled peanuts could offset the inconveniences caused by reactions in phase 2.

At the same time, the findings of a French clinical study have shown that a skin patch might be effective against peanut allergies and also potentially against other food allergies. Based on the hypothesis that the allergic shock affecting the immune system can be lessened directly through the skin, the patch contains a high concentration of peanut proteins. The proteins are absorbed by the skin, without entering the bloodstream, thereby ensuring a progressive desensitization to peanuts.

Findings have shown that 83.3% of participants were able to increase the quantity of peanuts they could eat without having an allergic reaction. The United States Food and Drug Administration (FDA) dubs this patch a “therapeutic breakthrough”, which could be marketed as early as 2018. Trials are still underway to determine whether the findings are just as promising for milk and egg allergies.

This summer, Australian researchers shared the results of their clinical trials at the Murdoch Children’s Research Institute in Melbourne in 2013. The hypothesis driving this study was that administering probiotics with low doses of peanut proteins could increase allergic resistance. Four years later, studies have shown that more than 82% of children who participated in the study, that is, children with peanut allergies, can now eat peanuts just like children without the allergy.

This data implies that not only would this treatment provide a true, long-term tolerance, but also that peanut allergies might be completely cured. The last clinical trials to corroborate this data will soon be underway to market a product with these kinds of therapeutic benefits.

Pilot project in the works at Sainte-Justine UHC

Oral immunotherapy, which is more accessible, might also enable the immune system to limit its excessive reactions to food allergens. This is done by exposing the body to microdoses (0.1 milligrams, or the equivalent of 1/2,500 of a peanut) and increasing them gradually. Results seem largely positive, with a marked tolerance of 80% for peanuts, milk and eggs. A pilot project authorized by Gaétan Barrette, Health and Social Services Minister, has just been launched at the Sainte-Justine UHC where they intend to desensitize 225 patients from 2017 to 2018, and 275 per year for the next two years.

These promising treatments bring into question the preconceived notion that allergies “cannot be cured” and that “you have to live with it”. Knowing that an allergy can happen at any time in life and not only in childhood, these findings give hope to thousands of people suffering from food allergies.

31 Oct 2017  |  Written by :

Julie Ruel, manager, Taxation, Raymond Chabot Grant Thornton

See the profile

Next article

Update on Measures Proposed by Federal Government to Abolish Certain Tax Planning Using Private Corporations, Tax Reductions Announced for SMEs and Other Measures

On October 16, 2017, the federal government issued a release in which it signaled its intention to simplify the measures proposed on July 18, 2017 to counter family income splitting. Furthermore, the government confirms that it will not be adopting the measures proposed for restricting access to the lifetime capital gains exemption (LCGE). In a new release issued October 19, 2017, the Finance Minister of Canada added that the measures relating to the conversion of income into capital gains will not be implemented. Concurrently with these announcements, the government communicated its intention to reduce the tax rate for SMEs as of 2018.

Lastly, in its October 24, 2017 Economic Statement, the federal government reiterated all of the recent days announcements, adding two measures for families and low-income individuals, i.e. indexation of the Canada Child Benefit as of 2018 and enhancement of the Working Income Tax Benefit.

Here is a brief summary of the tax measures announced by the federal government since October 16, 2017.

Next article

Does your SR&ED project qualify for claims?

Here is a second caselaw analysis.

We published the first article in the April edition of Strangely Techie in which we analyzed an SR&ED case at the Tax Court of Canada. Here is a second one dealing with a slightly different issue: Flavor Net Inc. v. The Queen, recently settled in September 2017.

Two SR&ED projects rejected

In this case, the taxpayer contested the CRA’s rejection of two of its projects.

Project #1 consisted in the development of an energy drink containing 800 milligrams of plant sterols uniformly dispersed in a 2-oz. volume, while existing commercial products do not exceed one eighth of this concentration. The challenge lay in the fact that plant sterols are hydrophobic substances that do not mix well in a water-based solution.

Project #2 had been described in the T661 form as the development of a procedure for filling a dual-chambered bottle. However, during the hearing, this project was instead presented as the development of a partial hot-fill system in which only the active component is pasteurized and later diluted in the sterile, distilled water.

Non-eligible SR&ED project: causes for rejection

The Judge rejected the eligibility of Project #1 on the basis of five currently used criteria. In particular, he rendered his decision on the absence of technological uncertainty, because the business contented itself to use a combination of solutions and methods that are well established in the food industry: intensive mixing devices, heat, emulsifiers and dispersion ingredients. The Judge’s ruling was therefore that this was a project completed by “routine engineering”. The purpose of the project was obtained using well-known techniques, with a reasonably predictable result. The development of a new product does not necessarily require the resolution of technological uncertainty. The fact that a product does not exist on the market is not sufficient to demonstrate the presence of SR&ED.

The Judge also commented on the fact that the taxpayer had not shown that he had established the state of knowledge currently accessible at the beginning of the project. In other words, an evaluation of the technological status (search for existing patents, overview of the technical documents) must be performed before starting, and all evidence kept.

While the previous element was enough to dispose of the appeal with Project #1, the decision identified other faults, in particular, the failure to follow a scientific method, which includes the formulation of hypotheses and the realization of trials to verify these hypotheses. A hypothesis must be specific and not simply the rewording of the project’s objective as a question. “Trial and error” work is not considered respecting the scientific process; it is an approach that can be defined as a series of trials performed without following a clear, iterative logical process.

Furthermore, the Judge examined the last criteria, which was documentation, and concluded that it was very thorough. He also mentioned that while documentation was not compulsory, it assists the taxpayer in establishing that his activities do qualify as SR&ED.


As for Project #2, the Judge confirmed his rejection on the basis of confusion and inconsistency in the case. First, the project content described during the hearing and during the CRA’s review, does not correspond to what had been submitted in the T661 form. Second, the documentation did not substantiate whether the work was performed in the year in which it is being claimed. This inconsistency and absence of corroboration contributed to the project being rejected.

To be noted

In short, the lessons we can learn from this case are as follows:

  • An SR&ED project should stand out from routine engineering even if the work is long and complex. Routine engineering is defined as the use of current techniques that yield a “reasonably predictable” result. However, according to previous decisions, if the result of non-trivial combinations of standard practices cannot be predicted by experts in the field, we may be in the presence of SR&ED.
  • The state of technological knowledge must be well established before starting the project, and must be documented.
  • SR&ED project experiments must be justified based on scientific and novel hypotheses. The reasons behind each trial are what separate systematic investigation from trial and error.
  • It is important to file clear, thorough and convincing claims. In some cases, this can be difficult when you have a maximum of 1,400 words for the technical description of a project. Furthermore, information provided should be consistent with the documentation available.