Faraz Malik Awan is working as a Research Engineer at Data Intelligence and Communication Engineering (DICE) Lab, Telecom SudParis. This job is in collaboration with his Ph.D. from Institut Polytechnique de Paris. He is conducting research on recommendation/prediction systems for Smart Cities, exploiting Machine/Deep Learning approaches. He did his BS in Computer Science from COMSATS University Islamabad. After completing his BS degree, he started working for a German corporation, Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) as a Software Engineer.

Almost 200 years ago, the renowned U.S. naturalist John James Audubon hid a decaying pig carcass under a pile of brush to test vultures' sense of smell. When the birds overlooked the pig—while one flocked to a nearly odorless stuffed deer skin—he took it as proof that they rely on vision, not smell, to find their food. His experiment cemented a commonly held idea. Despite later evidence that vultures and a few specialized avian hunters use odors after all, the dogma that most birds aren't attuned to smell endured. Now, that dogma is being eroded by findings on birds' behavior and molecular hardware, two of which were published just last month. One showed storks home in on the smell of freshly mowed grass; another documented scores of functional olfactory receptors in multiple bird species. Researchers are realizing, says evolutionary biologist Scott Edwards of Harvard University, that “olfaction has a lot of impact on different aspects of bird biology.” Forty years ago, when ethologist Floriano Papi proposed that homing pigeons find their way back to a roost by sniffing out its chemical signature, his colleagues scoffed at the idea. They pointed out that birds have several other keen senses to guide them, including sight and, in the case of pigeons and some other species, a magnetic sense. “By then, biological textbooks already stated unequivocally that birds have little to no sense of smell, and many people still believe it—even scientists,” says Danielle Whittaker, a chemical ecologist at Michigan State University. Still, contrary evidence was already accumulating. In the 1960s, ornithologist Kenneth Stager found vultures were attracted to boxes with a carcass hidden inside and fans that vented the odors—as long as this bait wasn't too decomposed, as was likely the case in Audubon's experiment. Researchers also found that albatrosses, shearwaters, and some other seabirds find their fish prey by detecting a chemical released by the plankton the fish eat. But these birds, forced to navigate many kilometers across a featureless sea, seemed exceptional. In 2008, “You were part of the dark side if you talked about birds using olfaction,” recalls Martin Wikelski, an ecologist at the Max Planck Institute for Ornithology. That year, though, a graduate student at his institute, molecular ecologist Silke Steiger, analyzed nine bird genomes from across the avian family tree and uncovered many genes for olfactory receptors—proteins in the nasal passages that bind to odors and relay a signal to the brain. In species that don't rely much on smell (humans are an example), these genes often mutate and become nonfunctional. But the researchers confirmed that many of the birds' olfactory genes were intact. What's more, they found that the number of these genes correlated with the size of a bird species' olfactory bulb, the brain's smell center—further evidence that the receptors were functional. The genomes in that study were incomplete, however. Last month, Christopher Balakrishnan, an evolutionary biologist at East Carolina University, and graduate student Robert Driver examined some of the best available bird genomes and for some species found many more olfactory genes. Their analysis of genomes from a hummingbird, emu, chicken, zebra finch, and a tropical fruit eater called a manakin revealed scores of new olfactory receptors, they reported on 28 June in the journal Integrative and Comparative Biology . That the emu has so many of these genes excites Whittaker, because this bird sits near the base of the bird family tree. “This result suggests that the ancestor to all birds must have had a very diverse set of olfactory receptor genes as well,” she says. Smell must have been important to birds from the beginning, and comparisons of their olfactory receptor genes today confirm it remains so. Balakrishnan and Driver found that one diverse set of receptors unique to birds has split into multiple types specific to different bird lineages. That suggests these genes evolved rapidly as the birds diversified. Natural selection may have honed the genes to perform crucial tasks. Wikelski and colleagues saw bird smell in action after they were inspired by a question from a curious primary school student. During an outreach program at a school in Radolfzell, Germany, the student asked the scientists how the local population of European white storks found their way to freshly cut meadows, where their insect and rodent prey were most exposed. To find out, Wikelski piloted his plane in circles to observe a flock of 70 storks on sunny spring and summer days. Even when the storks couldn't see or hear the mowing, he and his colleagues noted, they homed in on mowed fields upwind of them, as if drawn to the smell of the cut grass. To confirm the suspicion, the team sprayed cut-grass smell—a mix of three volatile chemicals—onto fields that hadn't been mowed recently. The storks came flocking, the team reported on 18 June in Scientific Reports . The work “shows very clearly that these birds rely exclusively on their sense of smell to make foraging decisions,” Whittaker says. Other bird species may also respond to “calls” from injured plants, recent evidence shows. Two European birds, the great tit and the blue tit, locate insects that are attacking pine trees by detecting the volatile chemicals the stressed trees release, ecologist Elina Mäntylä of the Biology Centre of the Czech Academy of Sciences and colleagues reported in the September 2020 issue of Ecology and Evolution . All these results show bird olfaction “should not be ignored,” Mäntylä says. Driver adds that they might also point to a new form of natural pest control, in which farmers or foresters could treat threatened flora with chemicals that entice birds to come and gobble up invasive insects. Other studies suggest olfaction might guide social interactions between birds. Whittaker's team has focused on preen oil, which birds secrete from a gland at the base of the tail and rub onto their feathers. The oil's chemical composition reveals the bird's species, sex, aggressiveness, and reproductive state. Females produce much more of these odorous chemicals, Whittaker and her colleagues reported in January in the Journal of Chemical Ecology , suggesting they depend more on odors to communicate, lacking the flashy feathers and songs that males rely on. Use of these cues is “likely widespread,” says Steiger, now at the German chemical company BASF SE, “but simply not yet investigated well enough.” That's changing fast, as studies of bird olfaction expand into new species. Published papers on the topic have doubled every decade since 1992, reaching 80 this past year. The field is, belatedly, putting Audubon's misconception to rest and acknowledging that birds—champions of flight, vision, and song—have another power as well.

The world is now at a crucial point in technology evolution. The future depends on its deployment across all industries today, including in higher education. Artificial Intelligence-based technologies are thought to promote rather than hinder democratic values including freedom, equality, and transparency. AI-based technologies can become a tool to promote equity and personalized learning. For the past 20 years, Artificial Intelligence (AI) has made some advances in higher education, but not enough.

As artificial intelligence (AI) gains a foothold, your next favorite teacher may be a silicon-based bag of bolts. We all remember a favorite teacher. But I wonder if what we recall is the social engagement or the actual teaching ability. I would imagine that it's a combination of both. Yesterday's and today's great teachers have that knack for bringing information to life and igniting a desire to learn.

Would you like to learn a language from a friendly piece of software? Berlitz has been in the business of language instruction for 143 years. They've been looking to move forward with 21st Century technology, but at the same time "Berlitz has built our methodology and brand on delivering the best outcomes for students serious about fluency, which requires a very human centric experience," said Curt Uehlein, Berlitz's CEO. Digital online resources for learning language are surprisingly sparse. Search for language instruction videos on YouTube and you'll find a handful of short, often homemade, videos covering some basic vocabulary.

I hope this was useful, don't hesitate to reach out to me via LinkedIn if you have strong opinions about this process. Also, if you want to learn more about a specific machine learning topic, take a look at my Youtube channel. In this section I'll share a collection of learning resources I recommend for people wanting to get started learning. This is not an exhaustive list, but it will be a good starting point for people wanting to get a good first mental model of Machine Learning. A classic in the machine learning community, highly recommend spending time with this book over and over again.

Natural language inference (NLI) is known as one of the central tasks in natural language processing (NLP) which encapsulates many fundamental aspects of language understanding. With the considerable achievements of data-hungry deep learning methods in NLP tasks, a great amount of effort has been devoted to develop more diverse datasets for different languages. In this paper, we present a new dataset for the NLI task in the Persian language, also known as Farsi, which is one of the dominant languages in the Middle East. This dataset, named FarsTail, includes 10,367 samples which are provided in both the Persian language as well as the indexed format to be useful for non-Persian researchers. The samples are generated from 3,539 multiple-choice questions with the least amount of annotator interventions in a way similar to the SciTail dataset. A carefully designed multi-step process is adopted to ensure the quality of the dataset. We also present the results of traditional and state-of-the-art methods on FarsTail including different embedding methods such as word2vec, fastText, ELMo, BERT, and LASER, as well as different modeling approaches such as DecompAtt, ESIM, HBMP, and ULMFiT to provide a solid baseline for the future research. The best obtained test accuracy is 83.38% which shows that there is a big room for improving the current methods to be useful for real-world NLP applications in different languages. We also investigate the extent to which the models exploit superficial clues, also known as dataset biases, in FarsTail, and partition the test set into easy and hard subsets according to the success of biased models. The dataset is available at https://github.com/dml-qom/FarsTail

Gaussian processes with derivative information are useful in many settings where derivative information is available, including numerous Bayesian optimization and regression tasks that arise in the natural sciences. Incorporating derivative observations, however, comes with a dominating $O(N^3D^3)$ computational cost when training on $N$ points in $D$ input dimensions. This is intractable for even moderately sized problems. While recent work has addressed this intractability in the low-$D$ setting, the high-$N$, high-$D$ setting is still unexplored and of great value, particularly as machine learning problems increasingly become high dimensional. In this paper, we introduce methods to achieve fully scalable Gaussian process regression with derivatives using variational inference. Analogous to the use of inducing values to sparsify the labels of a training set, we introduce the concept of inducing directional derivatives to sparsify the partial derivative information of a training set. This enables us to construct a variational posterior that incorporates derivative information but whose size depends neither on the full dataset size $N$ nor the full dimensionality $D$. We demonstrate the full scalability of our approach on a variety of tasks, ranging from a high dimensional stellarator fusion regression task to training graph convolutional neural networks on Pubmed using Bayesian optimization. Surprisingly, we find that our approach can improve regression performance even in settings where only label data is available.

Many hierarchical reinforcement learning (RL) applications have empirically verified that incorporating prior knowledge in reward design improves convergence speed and practical performance. We attempt to quantify the computational benefits of hierarchical RL from a planning perspective under assumptions about the intermediate state and intermediate rewards frequently (but often implicitly) adopted in practice. Our approach reveals a trade-off between computational complexity and the pursuit of the shortest path in hierarchical planning: using intermediate rewards significantly reduces the computational complexity in finding a successful policy but does not guarantee to find the shortest path, whereas using sparse terminal rewards finds the shortest path at a significantly higher computational cost. We also corroborate our theoretical results with extensive experiments on the MiniGrid environments using Q-learning and other popular deep RL algorithms.

You're looking for a complete Machine Learning course that can help you launch a flourishing career in the field of Data Science & Machine Learning, right? You've found the right Machine Learning course! Check out the table of contents below to see what all Machine Learning models you are going to learn. How this course will help you? A Verifiable Certificate of Completion is presented to all students who undertake this Machine learning basics course.