Technology / Gadgets

Amazon takes on supermarkets with drive-through grocery pickup

Over the last couple of decades, Amazon has slowly conquered the online retail space. The Seattle-based e-commerce company has made it possible to purchase any thing from books to live ladybugs within seconds. But the site hasn't seen the same succes...

Amazon’s cashier-free store breaks when more than 20 people are shopping


Amazon announced its intention to put the nation's cashiers out of work a few months ago, but it looks like they'll be getting a reprieve for now.

The post Amazon’s cashier-free store breaks when more than 20 people are shopping appeared first on ExtremeTech.

Nintendo’s DSi Shop is somehow still open, but not for much longer

Nintendo's DSi portable console, the precursor to the 3DS, was released way back in early 2009. Amazingly enough, the DSi's online shop is still up and running, even though Nintendo stopped selling the console years ago. But those days are about to e...

The Roadie 2 gives you no excuse for an out-of-tune guitar

I have what has been described as a "really crappy guitar." It's not even mine. It belongs to my friend who, before handing it to me, said, "yeah, I let my kid hit this thing with stuff. Also, I've never changed the strings." It seemed like the perfe...

Watch this Periscope video after a short message from #brands

Video ads are nothing new on Twitter. The company has announced a number of ways for brands to get their message out through promoted tweets, pre-roll messages and even ads in Moments. Now Twitter is extending its advertising to live video. Starting...

Tinder on the desktop: Looking for love in another wrong place

Tinder has arrived to the desktop with Tinder Online, finally helping you find love (or sex) on the web without a smartphone or tablet. The dating site describes the web app as "your English professor's worst nightmare," letting you seek partners whe...

Elon Musk wants to implant an AI interface in your brain

Deep Brain Stimulation 2

Electric cars, trips to Mars, solar everything, underground tunnels -- and now apparently brain-computer interfaces? Elon Musk has had another bright idea.

The post Elon Musk wants to implant an AI interface in your brain appeared first on ExtremeTech.

Maker Spotlight: Julia Dvorin

Julia Dvorin wants makers from around the world to fly their Freak Flag high! She aims to highlight the diversity within the maker movement.

Read more on MAKE

The post Maker Spotlight: JuliaDvorin appeared first on Make: DIY Projects and Ideas for Makers.

How big, fast data is transforming business intelligence

O'Reilly Podcast: Ian Fyfe of Zoomdata on the importance of “speed-of-thought analysis” in modern data environments.

In this podcast episode, I speak with Ian Fyfe, senior director for product marketing at Zoomdata, about the next generation of business intelligence software and how it addresses what Fyfe calls “the modern world of big and streaming data.”

Fyfe lists three aspects of modern data that have strained traditional BI systems:

Big data: many companies need to quickly query billions of records, which is a scale that traditional relational databases weren’t designed to handle. High-velocity data streams: as more devices get connected to the internet, many applications will involve hundreds or thousands of records per second streaming into databases in real time. Unstructured data: relational databases were meant to handle well-structured data—integers, dates, times, short strings, and so on. Now companies need to analyze large blocks of rich unstructured data like social media postings, customer reviews, and call-center logs.

Moreover, customer demands have changed. Users of BI software have become more sophisticated, and expect to have data-exploration tools “infused into every application, every business process,” says Fyfe. And they want short feedback loops, with queries answered instantaneously—what Fyfe calls “speed-of-thought analysis” that enables real-time research. “You’re not going to use a tool if it takes more than a few seconds to run,” says Fyfe.

Fyfe also describes a couple of case studies:

A pharmaceutical company with billions of rows of data on patients who aren’t getting potentially helpful therapies. An auto insurer with a vast database of historical price quotes. In the past, querying the database took so long that the company only used data to inform 3% of its new price quotes, but faster analytics have made it possible to inform every new quote with historical insight.

The business intelligence market is generally mature, Fyfe says, but four factors are pushing it into a new generation of technology:

The increasing complexity of data sets, especially streaming data. The rise of embedded analytics, and the expectation that analytics and visualization will be infused into every application and business process. End users have become more sophisticated, and they expect to be able to conduct interactive data discovery. The cloud: “the business intelligence market has been a bit of a laggard in moving to the cloud,” says Fyfe, but new-generation BI software has become flexible as to whether it can run on-premise or in the cloud.

This post and podcast is a collaboration between O'Reilly and Zoomdata. See our statement of editorial independence.

Continue reading How big, fast data is transforming business intelligence.

The House votes today on stripping away your internet privacy

Security surveillance unlock privacy

Later today, the House of Representatives will vote on a measure to functionally kill internet privacy. If you don't want that to happen, this is your last shot to stop it.

The post The House votes today on stripping away your internet privacy appeared first on ExtremeTech.

This Pi-Powered Linux Terminal Is the Size of an iPhone

Making a handheld Linux device with the Raspberry Pi Zero W may seem a daunting task, but is easier than you think with the right materials.

Read more on MAKE

The post This Pi-Powered Linux Terminal Is the Size of aniPhone appeared first on Make: DIY Projects and Ideas for Makers.

Facebook Introduced Live Location in Messenger – Geek News Central

Facebook has introduced a new way to share your location in Messenger. The new Live Location feature makes it simple and seamless for you to choose to share where you are in the real world with your friends and family. This is rolling out globally and is available on both iOS and Android.

In a blog post, Facebook said:

We’ve been testing this for a little while, and people tell us that Live Location is helpful when trying to coordinate with friends, telling people how close you are when you’re on your way to an appointment, or even sharing where you are with your roommate when you’re on your way home at night. You can share your Live Location with a group of friends in Messenger or just with one person – it’s up to you!

The most important thing to know is that Messenger is not going to suddenly start sharing your Live Location with your friends. You have to actively click a few things when you want to share your Live Location with a friend (or friends).

Once you share your Live Location, the person or people you share it with will be able to see where you are on a map for the next 60 minutes. Users have the option to stop sharing their location at any time – by tapping a button called “Stop Sharing”.

This new feature also enables people to share a static point on a map (instead of sharing their Live Location). It is useful for when a group of people want to meet at a specific place. I can see people making use if this feature at crowded conferences or cafeterias.

The Live Location feature is still rolling out. You can check to see if it is available where you are by visiting the Facebook Help Center.  The last sentence on that post might say “Live Location isn’t available in your area yet.”

Facebook Introduced Live Location in Messenger

New advances in directed self-assembly could push silicon below 10nm more efficiently than EUV

Intel Foundry

A new paper claims that directed self-assembly could be key to patterning lines below 10nm, but line-edge roughness and manufacturing challenges still present substantial barriers.

The post New advances in directed self-assembly could push silicon below 10nm more efficiently than EUV appeared first on ExtremeTech.

Four short links: 28 March 2017

Injectable Electronics, Style Transfer, Face Recognition, and Statistical Post-Mortems

Syringe Injectable Electronics -- demonstrating syringe injection and subsequent unfolding of submicrometer-thick, centimeter-scale macroporous mesh electronics through needles with a diameter as small as 100 micrometers. Our results show that electronic components can be injected into man-made and biological cavities, as well as dense gels and tissue, with > 90% device yield. Deep Photo Style Transfer -- code and data for the paper on GitHub. FBI Face Recognition Database Woes (Guardian) -- Approximately half of adult Americans’ photographs are stored in facial recognition databases that can be accessed by the FBI, without their knowledge or consent, in the hunt for suspected criminals. About 80% of photos in the FBI’s network are non-criminal entries, including pictures from driver’s licenses and passports. The algorithms used to identify matches are inaccurate about 15% of the time, and are more likely to misidentify black people than white people. Digital Experimentation and Peer Effects -- there's this quote from Sir Ronald Fisher, pioneer in statistics, that to consult the statistician after the experiment has been conducted, is to merely ask him to do a post mortem. He can tell you what the experiment died of. So true! (Also true for female statisticians.)

Continue reading Four short links: 28 March 2017.

Adapting ideas from neuroscience for AI

Inspiration from the brain is extremely relevant to AI; it’s time we pushed it further.

A better understanding of the reasons why neurons spike could lead to smart AI systems that can store more information more efficiently, according to Geoff Hinton, who is often referred to as the “godfather” of deep learning.

Geoff Hinton is an emeritus distinguished professor at the University of Toronto and an engineering fellow at Google. He is one of the pioneers of neural networks, and was part of the small group of academics that nursed the technology through a period of tepid interest, funding, and development.

Key takeaways: Large-scale analysis of the brain through research schemes like the Obama administration’s “Brain initiative” have the promise to shed light on new aspects of the brain, giving AI designers new ideas. You can adapt ideas from neuroscience into ideas that are relevant to AI—though it takes some time. Hinton first thought, in 1973, about implementing a system with capabilities similar to those afforded by the fact synapses change on multiple timescales, yet it took until 2016 to publish a major paper on this area. It’s relatively easy to develop powerful perception systems, but we need new techniques to build systems capable of reasoning and language.

Jack Clark: Why should we look at the brain when developing AI systems, and what aspects should we focus on?

Geoff Hinton: The main reason is that it's the thing that works. It's the only thing we know that's really smart and has general purpose intelligence. The second reason is that, for many years, a subset of people thought you should look at the brain to try to make AI work better, and they didn't really get very far—they made a push in the 80s, but then it got stalled, and they were kind of laughed at by everyone in AI saying “you don't look at a bumblebee to design a 747.” But it turned out the inspiration they got from looking at the brain was extremely relevant, and without that, they probably wouldn't have gone in that direction. It's not just that we have an example of something that's intelligent; we also have an example of a methodology that worked, and I think we should push it further.

JC: Today, aspects of modern classifiers like neural nets look vaguely similar to what we know about the brain's visual system. We're also developing memory systems that are inspired by the hippocampus. Are there other areas we can look to the brain and start taking elements from, like spiking neurons?

GH: We don't really know why neurons spike. One theory is that they want to be noisy so as to regularize, because we have many more parameters than we have data points. The idea of dropout [a technique developed to help prevent overfitting] is that if you have noisy activations, you can afford to use a much bigger model. That might be why they spike, but we don't know. Another reason why they might spike is so they can use the analog dimension of time, to code a real value at the time of the spike. This theory has been around for 50 years, but no one knows if it's right. In certain subsystems, neurons definitely do that, like in judging the relative time of arrival of a signal to two ears so you can get the direction.

Another area is in the kinds of memory. Synapses adapt at many different timescales and in complicated ways. At present, in most artificial neural nets, we just have a timescale for adaptation of the synapses and then a timescale for activation of the neurons. We don't have all these intermediate timescales of synaptic adaptation, and I think those are going to be very important for short-term memory, partly because it gives you a much better short-term memory capacity.

JC: Are there any barriers to our ability to understand the brain that could slow down the rate at which we can develop ideas in AI inspired by it?

GH: I think if you stick an electrode in a cell and record from it, or put an electrode near a cell and record from it, or near a bunch of cells and try to record from half a dozen of them, then you won't understand things that might easily be understood by optical dyes, which let you know what a million cells are doing. There's going to be all sorts of things in the Obama initiative for brain science to give us new techniques that will allow us to see (and make obvious) things that would have been very hard to establish. We don't know what they're going to be, but I suspect that will lead to some interesting things.

JC: So, if we had a sufficiently large neural network, would that be able to match a human on any given task or are there missing components we need?

GH: It depends on what particular task you're talking about. If you take something like speech recognition, I'd be very surprised if a really big network exactly matched a human being; I think it's either going to be worse or it's going to be better. Human beings aren't the limit. I think actually in speech recognition, I wouldn't be at all surprised if in 10 year's time, neural nets can't do it better than people. For other areas, like reasoning and learning from a very small number of examples, it may take longer to develop systems that match or surpass people.

JC: One problem modern reinforcement learning systems seem to have is knowing what parts of a problem to devote attention to exploring, so you don’t have to waste your time on less interesting parts of the image.

GH: This is exactly the same in vision. People make very intelligent fixations. Almost all of the optical array never gets processed at high resolution, whereas in computer vision, people typically just take the whole array at low-resolution, medium-resolution, high-resolution, and try to combine the information, so it's just the same problem in us. How do you intelligently focus on things? We’re going to have to deal with the same problem in language. This is an essential problem, and we haven't solved it yet.

JC: You recently gave a lecture on a paper you published about short-term changes of weights within neural networks. Can you explain this paper and why you think it is important?

GH: In recurrent neural networks, if they're processing a sentence, they have to remember stuff about what has happened so far in the sentence, and all of that memory is in the activations in the hidden neurons. That means those neurons are having to be used to remember stuff, so they're not really available for doing current processing.

A good example of this is if you have an embedded sentence—like if someone said, “John didn't like Bill because he was rude to Mary, because Bill was rude to Mary”—you process the beginning of the sentence, then you use exactly the same knowledge processing to process “because Bill was rude to Mary.” Ideally, you want to use the same neurons and the same connections and the same weights for the connections for this processing. That's what true recursion would be, and that means you have to take what you have so far in a sentence and you have to put it aside somewhere. The question is: how do you put it aside? In a computer, it's easy because you have random access memory, so you just copy it into some other bit of memory to free up the memory. In the brain, I don't think we copy neural activity patterns; what I think we do is have rapid changes to synapse strength so we can recreate the memories when we need them, and we can recreate them when the context is such that it would be appropriate.

I have a recent paper with Jimmy Ba and some people at DeepMind showing how we can make that work. I think that's an example of where the fact that synapses are changing on multiple timescales can be useful. I first thought about this in 1973 and made a little model that could do true recursion on a very simple problem. A year ago, I went back to that at DeepMind and got it working within the framework so it learns everything. Back when I first thought about it, computers had 64k of memory and we didn't know how to train big neural nets.

JC: Do you think AI agents need to be embodied in some form, either in a robot or sufficiently rich simulation, to become truly intelligent?

GH: I think there are two aspects, one is the philosophical aspect and the other is the practical aspect. Philosophically, I see no reason why they have to be embodied, because I think you can read Wikipedia and understand how the world works. But as a practical matter, I think embodiment is a big help. There's a Marx phrase: “If you want to understand how the world works, try and change it.” Just looking is not as efficient a way of understanding things as acting. So, the philosophical question is: is action essential? If action is essential to understanding the world, then astrophysics is in trouble. So, no, I don’t think embodiment is necessary.

JC: If you’re able to replicate some of the properties of spiking neurons and combine that with systems that can form temporary memories, then what will you be able to build?

GH: I think it might just make all the stuff we have today work better. So, for natural language understanding, I think having an associative memory with fast changes in the weights would be helpful, and for these feedforward nets, I think coincidence detectors are much better at filtering out clutter in the background, so they'll be much better at focusing on the signal and filtering out the noise. This could also help with learning from small data sets.

Continue reading Adapting ideas from neuroscience for AI.

Caption this, with TensorFlow

How to build and train an image caption generator using a TensorFlow notebook.

Image caption generation models combine recent advances in computer vision and machine translation to produce realistic image captions using neural networks. Neural image caption models are trained to maximize the likelihood of producing a caption given an input image, and can be used to generate novel image descriptions. For example, the following are possible captions generated using a neural image caption generator trained on the MS COCO data set.

possible captions generated using a neural image caption generator Figure 1. Credit: Raul Puri, with images sourced from MS COCO data set.

In this article, we will walk through an intermediate-level tutorial on how to train an image caption generator on the Flickr30k data set using an adaptation of Google’s Show and Tell model. We use the TensorFlow framework to construct, train, and test our model because it’s relatively easy to use and has a growing online community.

Why caption generation?

Recent successes in applying deep neural networks to computer vision and natural language processing tasks have inspired AI researchers to explore new research opportunities at the intersection of these previously separate domains. Caption generation models have to balance an understanding of both visual cues and natural language.

The intersection of these two traditionally unrelated fields has the possibility to effect change on a wide scale. While there are some straightforward applications of this technology, such as generating summaries for YouTube videos or captioning unlabeled images, more creative applications can drastically improve the quality of life for a wide cross section of the population. Similar to how traditional computer vision seeks to make the world more accessible and understandable for computers, this technology has the potential to make our world more accessible and understandable for us humans. It can serve as a tour guide, and can even serve as a visual aid for daily life, such as in the case of the Horus wearable device from the Italian AI firm Eyra.

Some assembly required

Before we begin, we’ll need to do some housekeeping.

First, you will need to install Tensorflow. If this is your first time working with Tensorflow, we recommend that you first review the following article: Hello, TensorFlow! Building and training your first TensorFlow model.

You will need the pandas, opencv2, and Jupyter libraries to run the associated code. However, to simplify the install process we highly recommend that you follow the Docker install instructions on our associated GitHub repo.

You will also need to download the image embeddings and image captions for the Flickr30k data set. Download links are also provided on our GitHub repo.

Now, let's begin!

The image caption generation model image caption generation model Figure 2. Credit: Shannon Shih from Machine Learning at Berkeley. Horse image sourced from MS COCO.

At a high-level, this is the model we will be training. Each image will be encoded by a deep convolutional neural network into a 4,096 dimensional vector representation. A language generating RNN, or recurrent neural network, will then decode that representation sequentially into a natural language description.

Caption generation as an extension of image classification

Image classification is a computer vision task with a lot of history and many strong models behind it. Classification requires models that can piece together relevant visual information about the shapes and objects present in an image, to place that image into an object category. Machine learning models for other computer vision tasks such as object detection and image segmentation build on this by not only recognizing when information is present, but also by learning how to interpret 2D space, reconcile the two understandings, and determine where an object's information is distributed in the image. For caption generation, this raises two questions:

How can we build upon the success of image classification models, in retrieving important information from images? How can our model learn to reconcile an understanding of language, with an understanding of images? Leveraging transfer learning

We can take advantage of pre-existing models to help caption images. Transfer learning allows us to take the data transformations learned by neural networks trained on different tasks and apply them to our data. In our case, the VGG-16 image classification model takes in 224x224 pixel images and produces a 4,096 dimensional feature vector useful for categorizing images.

We can take the representation (known as the image embedding) from the VGG-16 model and use it to train the rest of our model. For the scope of this article, we have abstracted away the architecture of VGG-16 and have pre-computed the 4,096 dimensional features to speed up training.

Loading the VGG image features and image captions is relatively straightforward:

def get_data(annotation_path, feature_path): annotations = pd.read_table(annotation_path, sep='\t', header=None, names=['image', 'caption']) return np.load(feature_path,'r'), annotations['caption'].values Understanding captions

Now that we have an image representation, we need our model to learn to decode that representation into an understandable caption. Due to the serial nature of text, we leverage recurrence in an RNN/LSTM network (to learn more, read "Understanding LSTM Networks"). These networks are trained to predict the next word in a series given previous words and the image representation.

Long short-term memory (LSTM) cells allow the model to better select what information to use in the sequence of caption words, what to remember, and what information to forget. TensorFlow provides a wrapper function to generate an LSTM layer for a given input and output dimension.

To transform words into a fixed-length representation suitable for LSTM input, we use an embedding layer that learns to map words to 256 dimensional features (or word-embeddings). Word-embeddings help us represent our words as vectors, where similar word-vectors are semantically similar. To learn more about how word-embeddings capture the relationships between different words, check out "Capturing semantic meaning using deep learning."

In the VGG-16 image classifier, the convolutional layers extract a 4,096 dimensional representation to pass through a final softmax layer for classification. Because the LSTM cells expect 256 dimensional textual features as input, we need to translate the image representation into the representation used for the target captions. To do this, we utilize another embedding layer that learns to map the 4,096 dimensional image features into the space of 256 dimensional textual features.

Building and training the model

All together, this is what the Show and Tell Model looks like:

Show and Tell Model Figure 3. Source: "Show and Tell: Lessons learned from the 2015 MSCOCO Image Captioning Challenge."

In this diagram, {s0, s1, ..., sN} represent the words of the caption we are trying to predict and {wes0, wes1, ..., wesN-1} are the word embedding vectors for each word. The outputs {p1, p2, ..., pN} of the LSTM are probability distributions generated by the model for the next word in the sentence. The model is trained to minimize the negative sum of the log probabilities of each word.

def build_model(self): # declaring the placeholders for our extracted image feature vectors, our caption, and our mask # (describes how long our caption is with an array of 0/1 values of length `maxlen` img = tf.placeholder(tf.float32, [self.batch_size, self.dim_in]) caption_placeholder = tf.placeholder(tf.int32, [self.batch_size, self.n_lstm_steps]) mask = tf.placeholder(tf.float32, [self.batch_size, self.n_lstm_steps]) # getting an initial LSTM embedding from our image_imbedding image_embedding = tf.matmul(img, self.img_embedding) + self.img_embedding_bias # setting initial state of our LSTM state = self.lstm.zero_state(self.batch_size, dtype=tf.float32) total_ loss = 0.0 with tf.variable_scope("RNN"): for i in range(self.n_lstm_steps): if i > 0: #if this isn't the first iteration of our LSTM we need to get the word_embedding corresponding # to the (i-1)th word in our caption with tf.device("/cpu:0"): current_embedding = tf.nn.embedding_lookup(self.word_embedding, caption_placeholder[:,i-1]) + self.embedding_bias else: #if this is the first iteration of our LSTM we utilize the embedded image as our input current_embedding = image_embedding if i > 0: # allows us to reuse the LSTM tensor variable on each iteration tf.get_variable_scope().reuse_variables() out, state = self.lstm(current_embedding, state) print (out,self.word_encoding,self.word_encoding_bias) if i > 0: #get the one-hot representation of the next word in our caption labels = tf.expand_dims(caption_placeholder[:, i], 1) ix_range=tf.range(0, self.batch_size, 1) ixs = tf.expand_dims(ix_range, 1) concat = tf.concat([ixs, labels],1) onehot = tf.sparse_to_dense( concat, tf.stack([self.batch_size, self.n_words]), 1.0, 0.0) #perform a softmax classification to generate the next word in the caption logit = tf.matmul(out, self.word_encoding) + self.word_encoding_bias xentropy = tf.nn.softmax_cross_entropy_with_logits(logits=logit, labels=onehot) xentropy = xentropy * mask[:,i] loss = tf.reduce_sum(xentropy) total_loss += loss total_loss = total_loss / tf.reduce_sum(mask[:,1:]) return total_loss, img, caption_placeholder, mask Generating captions using inference

After training, we have a model that gives the probability of a word appearing next in a caption, given the image and all previous words. How can we use this to generate new captions?

The simplest approach is to take an input image and iteratively output the next most probable word, building up a single caption.

def build_generator(self, maxlen, batchsize=1): #same setup as `build_model` function img = tf.placeholder(tf.float32, [self.batch_size, self.dim_in]) image_embedding = tf.matmul(img, self.img_embedding) + self.img_embedding_bias state = self.lstm.zero_state(batchsize,dtype=tf.float32) #declare list to hold the words of our generated captions all_words = [] print (state,image_embedding,img) with tf.variable_scope("RNN"): # in the first iteration we have no previous word, so we directly pass in the image embedding # and set the `previous_word` to the embedding of the start token ([0]) for the future iterations output, state = self.lstm(image_embedding, state) previous_word = tf.nn.embedding_lookup(self.word_embedding, [0]) + self.embedding_bias for i in range(maxlen): tf.get_variable_scope().reuse_variables() out, state = self.lstm(previous_word, state) # get a one-hot word encoding from the output of the LSTM logit = tf.matmul(out, self.word_encoding) + self.word_encoding_bias best_word = tf.argmax(logit, 1) with tf.device("/cpu:0"): # get the embedding of the best_word to use as input to the next iteration of our LSTM previous_word = tf.nn.embedding_lookup(self.word_embedding, best_word) previous_word += self.embedding_bias all_words.append(best_word) return img, all_words

In many cases this works, but by "greedily" taking the most probable words, we may not end up with the most probable caption overall.

One possible way to circumvent this is by using a method called "Beam Search." The algorithm iteratively considers the set of the k best sentences up to length t as candidates to generate sentences of size t + 1, and keeps only the resulting best k of them. This allows one to explore a larger space of good captions while keeping inference computationally tractable. In the example below, the algorithm maintains a list of k = 2 candidate sentences shown by the path to each bold word for each vertical time step.

Beam Search method Figure 4.

Credit: Daniel Ricciardelli.

Limitations and discussion

The neural image caption generator gives a useful framework for learning to map from images to human-level image captions. By training on large numbers of image-caption pairs, the model learns to capture relevant semantic information from visual features.

However, with a static image, embedding our caption generator will focus on features of our images useful for image classification and not necessarily features useful for caption generation. To improve the amount of task-relevant information contained in each feature, we can train the image embedding model (the VGG-16 network used to encode features) as a piece of the caption generation model, allowing us to fine-tune the image encoder to better fit the role of generating captions.

Also, if we actually look closely at the captions generated, we notice that they are rather mundane and commonplace. Take this possible image-caption pair for instance:

image-caption pair Figure 5.

Credit: Raul Puri, with images sourced from the MS COCO data set.

This is most certainly a "giraffe standing next to a tree." However, if we look at other pictures, we will likely notice that it generates a caption of "a giraffe next to a tree" for any picture with a giraffe because giraffes in the training set often appear near trees.

Next steps

First, if you want to improve on the model explained here, take a look at Google's open source Show and Tell network, trainable with the MS COCO data set and an Inception-v3 image embedding.

Current state-of-the-art image captioning models include a visual attention mechanism, which allows the model to identify areas of interest in the image to selectively focus on when generating captions.

Also, if you are interested in this state-of-the-art implementation of caption generation, check out the following paper: Show, Attend, and Tell: Neural Image Caption Generation with Visual Attention.

Note: Don't forget to access the corresponding Python code and iPython notebooks for this article on GitHub!

This post is a collaboration between O'Reilly and TensorFlow. See our statement of editorial independence.

Continue reading Caption this, with TensorFlow.

Google Home and WiFi launch in the UK on April 6th

Google's Home speaker is coming to the UK on April 6th, the company said today. Hosting an event in a makeshift London home, the search giant unveiled its hardware plans for Europe, confirming that the smart speaker will cost £129, undercutting...

GNC #1188 Artificial Intelligence – Geek News Central

Artificial Intelligence has the opportunity to make us extinct if we are not careful. I look at how Elon Musk is in one hand scared of the technology and on the other investing and starting a company, who’s sole focus will to make all of us Hybrids so we can keep up with sentient artificial intelligence. Plus all the tech news of the day. It’s a packed show so listen now.

My New Personal YouTube Channel
Geek News Central Facebook Page.
Download the Audio Show File

Support my Show Sponsor:
30% off on New GoDaddy Orders cjcgnc30
$.99 for a New or Transferred .com cjcgnc99 @
$1.00 / mo Economy Hosting with a free domain. Promo Code: cjcgnc1hs
$1.00 / mo Managed WordPress Hosting with free Domain. Promo Code: cjcgncwp1

Show Notes:

Billions to Stop AI. Billions to make us AI. Facebook AI Leader? Some users are Idiots. *Evil Alarm App* FBI Face Recognition. Facebook Townhall. Prison Insanity. Mega Upload Battle continues. ISS Space Walks. NASA Engine Test. SpaceX Engine Test. Vector Space Systems. Jupiter is Weird. Samsung Note 7 Fate. Night Shift on Mac. Steady Cam with Drone. iPad Apple Tv Remote. WhatsApp Hammer. Hacking a Human Cell. ACT NOW Privacy Loss Eminent. Intel Optane. iPad Bomb Plot. 3d Printed Liver by 2020. Your iOS Update. Verizon 4k on its FIOS. Uber Self-Driving Car Crash. Windows 10 Upgrade. Windows 10 Class Action. Child and Water Heater Robot. Sumatra Bike Trail Entity Chased Down. Idiot Stunt.

GNC #1188 Artificial Intelligence

Trump executive order will dismantle Obama environmental regulations

On Tuesday, Trump will sign an executive order which aims to undo Obama's 2013 Climate Action Plan

IKEA launches its own low-cost smart lighting range

For many people, their first foray into the world of home automation begins with lighting. There's a good reason for this: smart bulbs easily fit into existing furnishings and can be operated using just a smartphone, which (mostly) everybody now owns...

Runways of the Future May Be Circular – Geek News Central

Circular RunwaySince the earliest days of powered aviation, runways have had a simple, consistent design: A long, straight path, usually made of concrete or asphalt (or sometimes nothing more than a dirt strip or open field). It seems logical enough, as aircraft need a certain amount of clearance to land or takeoff. Thousands of airports have been built around this runway concept. It seems like an unshakable foundation of modern aviation.

But one group is challenging the notion of the current standard runway. Their so-called Endless Runway project actually proposes circular runways, that would surround an airport terminal:

The fundamental principle of The Endless Runway is that the aircraft take-off and land on a large circular structure. This will allow for the unique characteristic that the runway can be used in any wind direction, thus making the runway independent of the direction of the wind and therefore also the airport capacity independent of the wind direction.

That somewhat technical explanation breaks down to this: If runways are laid out in a circular fashion, air traffic controllers could adjust the entry and exit points of aircraft depending on where ground conditions are most favorable within the circle. Modern runway designs are static and mostly immovable. If there are heavy crosswinds that could cause trouble for takeoffs and landings, pilots and controllers either have to make the best of a bad situation or potentially call for the runway to be closed. But the circular runway design means aircraft can be directed to use the safest part of the circle, no matter what the conditions may be on any other part of the circle.

This approach could make airports safer and more efficient. For more information on circular runways, check out this video recently produced by the BBC.

Runways of the Future May Be Circular

NASA prepares the ISS for its second space taxi dock

NASA has begun preparing the ISS' second space taxi dock for Boeing's CST-100 Starliner and SpaceX's Crew Dragon. Flight controllers at NASA Johnson Space Center have taken control of the station's robotic arm, the Canadarm 2, to relocate one of its...

Asian dust providing key nutrients for California’s giant sequoias

Dust from as far away as the Gobi Desert in Asia is providing more nutrients than previously thought for plants, including giant sequoias, in California's Sierra Nevada mountains, a team of scientists, including several from the University of California, Riverside, have found.

Entrepreneurs love their companies like parents love their children

A recent study shows that love is a major motivator both for parents and entrepreneurs. A multidisciplinary study asks whether entrepreneurs love their companies like parents love their children. The study used functional MRIs to study the brain activity of fathers and high-growth entrepreneurs. Fathers were shown pictures of their own children as well as other children they knew. Entrepreneurs were shown pictures of their own companies and other companies that they were familiar with.

Prostate screening often occurs without discussion of benefits, risks

Less than a third of men in a large national survey reported talking with their doctor about both the pros and cons of the PSA blood test for prostate cancer, and the likelihood has decreased further since a national panel recommended against the test.

Honesty may not be the best policy for hospital safety grades, study suggests

A new study finds that a well-known hospital grading system may put too much weight on the wrong things. The grades are based in part on hospitals' self-reported use of safety-related protocols. But the study show this had little in common with how a hospital did on independent measurements of hospital-acquired infections -- or with whether the government had penalized it for high infection or readmission rates.

New report lays plan to eliminate 90,000 hepatitis B and C deaths by 2030

Hepatitis B and C kill more than 20,000 people every year in the United States. A new report from the National Academies of Sciences, Engineering, and Medicine presents a strategy to eliminate these diseases as serious public health problems and prevent nearly 90,000 deaths by 2030.

Toward glow-in-the-dark tumors: New fluorescent probe could light up cancer

A fluorescent probe developed by Michigan Tech chemist Haiying Liu lights up the enzyme beta-galactosidase in a cell culture. The glowing probe-enzyme combination could make tumors fluoresce, allowing surgeons to cut away cancer while leaving healthy tissue intact.

SHSU study examines death penalty support in Mexico

In sharp contrast to previous studies of public support for the death penalty conducted in the US, Catholics in Mexico were found to be more likely to support capital punishment, while older Mexicans and those living in states that bordered the US were less likely to support the death penalty, according to researchers at Sam Houston State University.

The physics of wealth inequality

A Duke engineering professor has proposed an explanation for why the income disparity in America between the rich and poor continues to grow. According to the constructal law of physics, income inequality naturally grows along with the economy.

When writing interferes with hearing

A cochlear implant is an electronic device capable of restoring hearing in a profoundly deaf person by stimulating the nerve endings in the inner ear. However, results can be extremely variable. Using brain imaging techniques, a neuroscientist from University of Geneva and a Parisian ENT surgeon have managed to predict the success of a cochlear implant among people who became profoundly deaf in their adult life. This research may be found in Nature Communications.

Dementia: The right to rehabilitation

Rehabilitation is important for people with dementia as it is for people with physical disabilities, according to a leading dementia expert.

It’s not too late to conserve water resources in rapidly urbanizing areas

As climate change and population pressure intensify in suburbia, a new study by watershed scientist Timothy Randhir at UMass Amherst suggests that threats such as water shortages and poor quality can be met if managers begin to act now. 'A lot of studies in hydrology climate science focus on climate, but very few combine the two, land use and its synergy with climate change.' He modeled how this watershed is going to look in the next 90 years taking both into account.

How does oxygen get into a fuel cell?

In order for a fuel cell to work, it needs an oxidizing agent. TU Wien has now found a way to explain why oxygen does not always enter fuel cells effectively, rendering them unusable.

Mayo Clinic researchers identify interaction among proteins that cause cancer cells to metastasize

Researchers at Mayo Clinic have identified an interaction among proteins that allows cancer cells to grow and metastasize. They say the discovery may play a role in developing a better understanding of how tumors grow in a variety of malignancies, including breast, prostate, pancreatic, colon, lung and skin cancers.

Malaria parasites ‘walk through walls’ to infect humans

Researchers have identified proteins that enable deadly malaria parasites to 'walk through cell walls' -- a superpower that was revealed using the Institute's first insectary to grow human malaria parasites.The research has identified two parasite proteins that are the key to this superpower. The proteins could be targeted to develop much-needed antimalarial drugs or vaccines.

Information storage with a nanoscale twist

Discovery of a novel rotational force inside magnetic vortices makes it easier to design ultrahigh capacity disk drives.

Gold standards for nanoparticles

KAUST researchers reveal how small organic 'citrate' ions can stabilize gold nanoparticles, assisting research on the structures' potential.

Study: Dust helps regulate Sierra Nevada ecosystems

A new study released March 28 in the journal Nature Communications indicates it's important to understand how dust helps vegetation thrive, especially in light of the changing climate and land-use intensification.

New research disproves common assumption on cranial joints of alligators, birds, dinosaurs

Researchers from the University of Missouri School Of Medicine recently discovered that although alligators, birds and dinosaurs have a similar skull-joint shape, this does not guarantee that their movements are the same.

Marathon running may cause short-term kidney injury

According to a new Yale-led study, the physical stress of running a marathon can cause short-term kidney injury. Although kidneys of the examined runners fully recovered within two days post-marathon, the study raises questions concerning potential long-term impacts of this strenuous activity at a time when marathons are increasing in popularity.

Trump Action on Clean Power Plan threatens air quality, health, and economic benefits

'If we overturn the Clean Power Plan we will forfeit important health benefits and undermine the longstanding American tradition of energy innovation and clean air progress, at a time when we need it most.'

NASA sees Tropical Cyclone Debbie make landfall in Queensland

Tropical Cyclone Debbie made landfall in Queensland bringing heavy rainfall, hurricane-force winds, rough seas, and flooding. NASA-NOAA's Suomi NPP satellite provided a visible look at the storm from space while NASA's Aqua satellite analyzed cloud temperatures to determine the location of the strongest storms within.

Therapies that target dementia in early stages critical to success

Targeting dementia in the earlier stages of the condition could be critical for the success of future therapies, say researchers from the University of Bristol, who have found that the very earliest symptoms of dementia might be due to abnormal stability in brain cell connections rather than the death of brain tissue, which comes after.

A basis for the application of drought indices in China

The definition of a drought index is the foundation of drought research. It is of great importance to evaluate the applicability of drought indices in drought monitoring, forecasting systems and research. Performances of seven meteorological drought indices have been identified in China, using terrestrial water storage obtained from the Gravity Recovery and Climate Experiment, the observed soil moisture and streamflow.

Findings support use of less invasive hysterectomy for early-stage endometrial cancer

Researchers found similar rates of disease-free survival and no difference in overall survival among women who received a laparoscopic or abdominal total hysterectomy for stage I endometrial cancer, according to a study published by JAMA.

‘Flying syringes’ could detect emerging infectious diseases

Blood-sucking flies can act as 'flying syringes' to detect emerging infectious diseases in wild animals before they spread to humans, according to research published in the journal eLife.

Vitamin D, calcium supplementation among older women does not significantly reduce risk of cancer

Among healthy postmenopausal women, supplementation with vitamin D3 and calcium compared with placebo did not result in a significantly lower risk of cancer after four years, according to a study published by JAMA.

Desktop scanners can be hijacked to perpetrate cyberattacks

The researchers conducted several demonstrations to transmit a message into computers connected to a flatbed scanner. Using direct laser light sources up to a half-mile (900 meters) away, as well as on a drone outside their office building, the researchers successfully sent a message to trigger malware through the scanner. Watch a video of the drone attack

Chlamydia: How bacteria take over control

To survive in human cells, chlamydiae have a lot of tricks in store. Researchers of the University of Würzburg have now discovered that the bacterial pathogens also manipulate the cells' energy suppliers in the process.

Protein identified as potential druggable target for pancreatic cancer

A protein known as arginine methyltransferase 1 (PRMT1) may be a potential therapeutic target for pancreatic ductal adenocarcinoma (PDAC), the most common type of pancreatic cancer, and one of the most deadliest with a less than 10 percent, five-year survival rate. PRMT1 is involved in a number of genetic processes including gene transcription, DNA repair and signaling.

How bacteria hunt other bacteria

A bacterial species that hunts other bacteria has attracted interest as a potential antibiotic, but exactly how this predator tracks down its prey has not been clear. A Biophysical Journal study reveals that the bacterial predator Bdellovibrio bacteriovorus homes in on its target by taking advantage of fluid forces generated by its own swimming movements and those of its prey. These bring the bacteria in close proximity, giving BV a greater chance of successful attack.

Mustard seeds without mustard flavor: New robust oilseed crop can resist global warming

University of Copenhagen and the global player Bayer CropScience have successfully developed a new oilseed crop that is much more resistant to heat, drought and diseases than oilseed rape. The breakthrough is so big that it will feature as cover story of the April issue of Nature Biotechnology, the most prestigious journal for biotechnology research.

ASCO and Cancer Care Ontario update guideline on radiation therapy for prostate cancer

The American Society of Clinical Oncology (ASCO) and Cancer Care Ontario today issued a joint clinical practice guideline update on brachytherapy (internal radiation) for patients with prostate cancer. The update provides evidence-based recommendations for different patient risk groups, and specifies the most effective forms of brachytherapy.

Researchers find video games influence sexist attitudes

The images and roles of female characters in video games send a powerful message that can influence the underlying attitudes of gamers. Iowa State and French researchers found a link between video game exposure and sexism in a new study of more than 13,000 adolescents.

Forests fight global warming in ways more important than previously understood

Trees impact climate by regulating the exchange of water and energy between the Earth's surface and the atmosphere, an important influence that should be considered as policymakers contemplate efforts to conserve forested land, said the authors of an international study that appears in the journal Nature Climate Change.

Cornering endangered species

Geographic areas occupied by certain species shrink as they decline in abundance, leaving them more vulnerable to extinction by harvest.

Van Andel Research Institute installs cryo-EM to explore molecular basis of disease

Van Andel Research Institute (VARI) is now home to one of the world's most powerful microscopes -- one that images life's building blocks in startling clarity and equips VAI's growing team of scientists to push the limits of discovery in search of new treatments for diseases such as cancer and Parkinson's.

NASA’s James Webb space telescope completes acoustic and vibration tests

At NASA's Goddard Space Flight Center in Greenbelt, Maryland the James Webb Space Telescope team completed the acoustic and vibration portions of environmental testing on the telescope. These tests are merely two of the many that spacecraft and instruments endure to ensure they are fit for spaceflight.

Unraveling the functional diversity of longevity gene SIRT1

While the search for elixir of life has captivated human imagination for millennia, researchers around the world have put in efforts to extend healthy lifespan and reduce the burden of morbid diseases in an increasingly aging population. Researchers from the Tata Institute of Fundamental Research have now identified a control mechanism within a longevity gene, which is key to unraveling its functional diversity and is likely to boost efforts at designing specific pharmacological agents.

Interferon-beta producing stem cell-derived immune cell therapy on liver cancer

Induced pluripotent stem (iPS) cell-derived myeloid cells (iPS-ML) that produce the anti-tumor protein interferon-beta (IFN-beta) have been produced and analyzed by researchers from Kumamoto University, Japan. Using human iPS-ML in a mouse model, they found that the cells migrate to and deliver IFN-beta to liver tumors thereby reducing cancer proliferation and increasing survival time.

Lead exposure in childhood linked to lower IQ, lower status

A long-term study of 565 children who grew up in the era of leaded gasoline has shown that their exposure to the powerful neurotoxin may have led to a loss of intelligence and occupational standing by the time they reached age 38. Ninety-four percent of the children exceeded today's reference value of 5 micrograms of lead per deciliter of blood. For each 5-microgram increase in blood lead, a person lost about 1.5 IQ points.

Biologists find ‘skin-and-bones’ mechanism underlying zebrafish fin regeneration

University of Oregon biologists have figured out how zebrafish perfectly regenerate amputated fins with a precisely organized skeleton.

Abuse accelerates puberty in children

While it has long been known that maltreatment can affect a child's psychological development, new Penn State research indicates that the stress of abuse can impact the physical growth and maturation of adolescents as well.

New quantum gadget could make contactless payment more secure

A prototype gadget that sends secret keys to encrypt information passed from a mobile device to a payment terminal, could help to answer public concerns around the security of contactless and wireless transactions.In partnership with Nokia and Bay Photonics, Oxford University researchers from the Department of Physics, have devised a system for transmitting quantum keys that ensures data security.

Rogue breast tumor proteins point to potential drug therapies

For patients with difficult-to-treat cancers, doctors increasingly rely on genomic testing of tumors to identify errors in the DNA that indicate a tumor can be targeted by existing therapies. But this approach overlooks rogue proteins that may be driving cancer cells and also could be targeted with existing treatments, according to research led by Washington University School of Medicine in St. Louis.

Tomographic imaging advances considered good yet can lead to overdiagnosis in PE patients

Although advances in tomographic imaging have improved the sensitivity of ventilation-perfusion (V/Q) scans for pulmonary embolism (PE), they may lead to overdiagnosis by revealing small and clinically insignificant PEs, according to a state-of-the-art review published in the March 2017 issue of the American Journal of Roentgenology (AJR).

Evaluation between maternal mental health and discharge readiness

New research indicates that mothers with a history of mental health disorders feel less ready for discharge from the NICU than with mothers without a mental health history.

Insurance coverage for IVF increases chance of having baby

Women who pursue in vitro fertilization (IVF) to become pregnant are more likely to give birth if they have health insurance that covers the procedure, according to new research at Washington University School of Medicine in St. Louis. The key reason is financial rather than medical: For many people, the high cost for one IVF procedure prohibits women from seeking a second treatment if the first attempt fails. The study is published March 28 in The Journal of the American Medical Association.

Which self-help websites actually improve health? New research yields a list

From depression to weight loss, insomnia to cutting back on alcohol or cigarettes, the Internet teems with sites that promise to help people improve their health. Which of these really help -- with evidence from gold-standard studies to back up these claims? A new paper compiles only the best of the best: a list of over 40 sites backed by evidence from randomized controlled trials.

NIST physicists show ion pairs perform enhanced ‘spooky action’

Adding to strong recent demonstrations that particles of light perform what Einstein called 'spooky action at a distance,' in which two separated objects can have a connection that exceeds everyday experience, physicists at the National Institute of Standards and Technology (NIST) have confirmed that particles of matter can act really spooky too.

A molecular on/off switch for CRISPR

TSRI scientists reveal how viruses disable bacterial immune systems.

Scientist pioneers technology new to MSU, maps giant virus

In a laboratory at Michigan State University, scientists took a DIY approach to build a retrofitted cryo-electron microscope that allowed them to map a giant Samba virus -- one of the world's largest viruses.

Neurological diseases cost the US Nearly $800 billion per year

A new paper published in the Annals of Neurology reports the most common neurological diseases pose a serious annual financial burden for the nation.

Using a method from Wall Street to track slow slipping of Earth’s crust

An indicator for stock prices can be used with GPS data to automatically detect slow-slip earthquakes from a single station's observations, offering a new way to monitor seismic activity.

New type of sensor material developed

Hokkaido University scientists have succeeded in developing a nickel complex that changes color and magnetism when exposed to methanol vapor. The new material can potentially be used not only as a chemical sensor, but also with future rewritable memory devices.

Women with insurance coverage for IVF more likely to have live birth

Women with insurance coverage for in vitro fertilization (IVF) were more likely to attempt IVF again and had a higher probability of live birth than women who self-paid for IVF, according to a study published by JAMA.

Meet your new electronic trauma intervention

The popular building-block computer game Tetris might be more than an idle pastime that keeps you glued to a screen. Playing it shortly after experiencing a traumatic event seems to block some of the recurrent intrusive memories that people are often left with. The proof-of-concept of the role, which Tetris could play within psychological interventions after trauma, is described in Springer Nature's journal Molecular Psychiatry, in a joint study by Lalitha Iyadurai and Emily Holmes

Female menstrual cycle in a dish

What if women could have a miniature, personalized reproductive system made with their own tissues that could predict how they would respond to certain medications? Northwestern has developed the first phase of this technology, made with human tissue, which could eventually change the future of research and treatment of diseases in women's reproductive organs. It will allow physicians to test drugs for safety and effectiveness and better understand such diseases as endometriosis, fibroids, cancer and infertility.

It is easier for a DNA knot…

How can long DNA filaments, which have convoluted and highly knotted structure, manage to pass through the tiny pores of biological systems? This is the fascinating question addressed by Antonio Suma and Cristian Micheletti, researchers at the International School for Advanced Studies (SISSA) in Trieste who used computer simulations to investigate the options available to the genetic material in such situations. The study has just been published in PNAS.

Sharing expert experimental knowledge to expedite design

A new repository of metabolic information provides a quick tool for designing useful synthetic biological systems.

Parents who play Pokémon GO with kids: ‘It wasn’t really about the Pokémon’

In the first study to survey and interview parents who play 'Pokémon GO' with their children, families report a number of side benefits, including increased exercise, more time spent outdoors and opportunities for family bonding. However, some guilt about screen time persisted.

Why does the same exercise exert effects on individuals differently?

Selenoprotein P, a kind of hepatokine hormone secreted from the liver, has been found, through experiments with cultured muscle cells and mice and through clinical studies, to cause pathology named 'exercise resistance,' which prevents health promotion by physical exercise. The present results elucidate one of the reasons why individual responsiveness to exercise differs markedly as well as shed lights on development of therapy for lifestyle diseases due to lack of exercise, obesity and type-2 diabetes.

Tiny sensor lays groundwork for precision X-rays detection via endoscopy

Using a tiny device known as an optical antenna, researchers have created an X-ray sensor that is integrated onto the end of an optical fiber just a few tens of microns in diameter. By detecting X-rays at an extremely small spatial scale, the sensor could be combined with X-ray delivering technologies to enable high-precision medical imaging and therapeutic applications.

Evidence insufficient to screen for celiac disease

The US Preventive Services Task Force has concluded that the current evidence is insufficient to assess the balance of benefits and harms of screening for celiac disease in asymptomatic persons. The report appears in the March 28 issue of JAMA.

How do we measure temperature? (video)

How do the thermometers in the kitchen or the doctor's office work? Thanks to the laws of thermodynamics, thermometers respond to heat moving from hot to cold as a means of measuring temperature. Clever physical chemists and engineers and even Einstein have made thermometers from a variety of materials. Watch the latest Reactions video here:

New report finds EPA’s controlled human exposure studies of air pollution are warranted

A new report by the National Academies of Sciences, Engineering, and Medicine finds these studies are warranted and recommends that they continue under two conditions: when they provide additional knowledge that informs policy decisions and regulation of pollutants that cannot be obtained by other means, and when it is reasonably predictable that the risks for study participants will not exceed biomarker or physiologic responses that are of short duration and reversible.

Hair spacing keeps honeybees clean during pollination

A honeybee can carry up to 30 percent of its body weight in pollen because of the strategic spacing of its nearly three million hairs. The gap between each eye hair is approximately the same size as a grain of dandelion pollen, which is typically collected by bees. This keeps the pollen suspended above the eye and allows the forelegs to comb through and collect the particles.

5,000 and counting: Penn Medicine celebrates reconstruction milestone

Doctors in Penn Medicine's Division of Plastic Surgery recently performed their 5,000th free flap reconstructive surgery. Their milestone will be the focus of a presentation at the 96th Annual American Association of Plastic Surgeons Meeting in Austin, Texas, in which they will educate leaders from other institutions on the process of building a free flap program.

About time! Predicting midge seasonality key to reducing livestock diseases

Ecologists at the UK-based Centre for Ecology & Hydrology (CEH) have led a study which informs optimal strategies for control of devastating midge-borne diseases like bluetongue and Schmallenberg virus that affect cattle and sheep in the UK and beyond.

Biomechanical analysis of head injury in pediatric patients

The biomechanics of head injury in youths (5 to 18 years of age) have been poorly understood. A new study reported in the Journal of Neurosurgery: Pediatrics set out to determine what biomechanical characteristics predispose youths with concussions to experience transient or persistent postconcussion symptoms.

To be or not to be … An entrepreneur

Prof. Ross Levine found evidence that a company's legal status -- incorporated or unincorporated -- can be used as a reliable measure to distinguish entrepreneurs from other business owners.

Childhood lead exposure associated with lower IQ, socioeconomic status nearly 3 decades later

Children who had higher blood lead levels at age 11 were more likely to have lower cognitive function, IQ and socioeconomic status when they were adults at age 38, according to a study published by JAMA.

Case Western Reserve University researchers turn urine into research tools

Researchers from Case Western Reserve University School of Medicine have developed a breakthrough technique to harvest cells directly from urine, and grow them into durable, clinically relevant stem cells to study Down syndrome.

Impacts of school choice on segregation

Diversity in schools is important for students' experiences and outcomes in schools and beyond, reducing prejudices and ensuring the likelihood of living and working in integrated environments as adults. Penn State researchers are exploring how school choice is affecting racial composition and segregation in Pennsylvania schools.

Brain stimulation improves schizophrenia-like cognitive problems

A new study from the University of Iowa Carver College of Medicine finds that stimulating the cerebellum in rats with schizophrenia-like thinking problems normalizes brain activity in the frontal cortex and corrects the rats' ability to estimate the passage of time -- a cognitive deficit that is characteristic in people with schizophrenia.

Astaxanthin compound found to switch on the FOX03 ‘Longevity Gene’ in mice

An Astaxanthin compound has been found to switch on the FOX03 'Longevity Gene' in a study using mice at the University of Hawaii. Researchers measured a nearly 90% increase in the activation of the gene in the animals' heart tissue. Life sciences company Cardax, Inc. looks forward to further confirmation in human clinical trials of Astanxanthin's potential role as an anti-aging therapy.

Physics can predict wealth inequality

The 2016 election year highlighted the growing problem of wealth inequality and finding ways to help the people who are falling behind. This human urge of compassion isn't new, but the big question that remains to be addressed is why inequality is so difficult to erase. This inspired Adrian Bejan at Duke University, who in 1996 discovered the Constructal Law, to provide an answer.

Novel approach can reveal personalized breast cancer treatments

Researchers from various institutions, including Baylor College of Medicine, have developed a new way to approach breast cancer treatment. First, they search for the proteins that drive tumor growth, and then test in the lab drugs that potentially neutralize these specific biological drivers.

Researchers uncover secret of nanomaterial that makes harvesting sunlight easier

Using sunlight to drive chemical reactions, such as artificial photosynthesis, could soon become much more efficient thanks to nanomaterials.
Go to Top

Hit Counter provided by Seo Australia