1. Institute of Neurobiology, Eberhard Karls University of Tübingen, Tübingen, Germany;
2. Werner-Reichardt Centre for Integrative Neuroscience, Tübingen, Germany;
3. Graduate Training Centre of Neuroscience – International Max-Planck Research School (IMPRS), Tübingen, Germany;
4. Chinese Academy of Sciences, Key Laboratory of Brain Connectome and Manipulation, The Brain Cognition and Brain Disease Institute, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Nanshan, China;
5. Shenzhen-Hong Kong Institute of Brain Science-Shenzhen Fundamental Research Institutions, Shenzhen, China

Download

manuscript pdf coming soon…

Abstract

Editor’s evaluation

This study presents a major technical advance by recording from genetically identifed neurons in freely moving mice. This method is applied to the hippocampus to determine circuit specifc synaptic interaction in vivo and to compare behavioral correlates of genetically-defned cell types. This technique paves the way for future studies aiming to relate genetics, circuits, and neuronal coding in freely moving animals.

Acknowledgments

Loading more memories in Burgalossi’s lab…

1. Guangdong Provincial Key Laboratory of Brain Connectome and Behavior, CAS Key Laboratory of Brain Connectome and Manipulation, the Brain Cognition and Brain Disease Institute (BCBDI), Shenzhen; Key Laboratory of Translational Research for Brain Diseases, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen-Hong Kong; Institute of Brain Science-Shenzhen Fundamental Research Institutions, Shenzhen 518055, China

Download

screenshot from Cerebral cortex webpage

manuscript pdf coming soon…

Abstract

In the process of information transmission, information is thought to be transmitted from the networks that are activated by the input to the networks that are silent or nonactivated. Here, via numerical simulation of a 3-network motif, we show that the silent neuronal network when interconnected with other 2 networks can exert much stronger causal influences on the other networks. Such an unexpected causal relationship results from high degree of synchronization in this network. The predominant party is consistently the network whose noise is smaller when the noise level in each network is considered. Our results can shed lights on how the internal network dynamics can affect the information flow of interconnected neuronal networks.

Acknowledgments

Loading wonderful memories in Zhan yang’s lab…

1. Guangdong Provincial Key Laboratory of Brain Connectome and Behavior, CAS Key Laboratory of Brain Connectome and Manipulation, the Brain Cognition and Brain Disease Institute (BCBDI), Shenzhen; Key Laboratory of Translational Research for Brain Diseases, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen-Hong Kong; Institute of Brain Science-Shenzhen Fundamental Research Institutions, Shenzhen 518055, China
2. State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang 110016, China
3. Institutes for Robotics and Intelligent Manufacturing, Chinese Academy of Sciences, Shenyang 110169, China
4. University of Chinese Academy of Sciences, Beijing 100049, China
5. Institute of Neuroinformatics, the University of Zürich and Swiss Federal Institute of Technology (ETH), Zurich CH-8057, Switzerland
6. European Molecular Biology Laboratory (EMBL), Monterotondo 00015, Italy
7. School of Systems Science, Beijing Normal University, Beijing 100875, China

Download

screenshot from Scibulletin webpage

manuscript pdf available from researchgate webpage

Abstract

During free exploration, the emergence of patterned and sequential behavioral responses to an unknown environment reflects exploration traits and adaptation. However, the behavioral dynamics and neural substrates underlying the exploratory behavior remain poorly understood. We developed computational tools to quantify the exploratory behavior and performed in vivo electrophysiological recordings in a large arena in which mice made sequential excursions into unknown territory. Occupancy entropy was calculated to characterize the cumulative and moment-to-moment behavioral dynamics in explored and unexplored territories. Local field potential analysis revealed that the theta activity in the dorsal hippocampus (dHPC) was highly correlated with the occupancy entropy. Individual dHPC and prefrontal cortex (PFC) oscillatory activities could classify various aspects of free exploration. Initiation of exploration was accompanied by a coordinated decrease and increase in theta activity in PFC and dHPC, respectively. Our results indicate that dHPC and PFC work synergistically in shaping free exploration by modulating exploratory traits during emergence and visits to an unknown environment.

Acknowledgments

Loading wonderful memories in Zhan yang’s lab…

Abstract

(1) Introduce topics, keywords, and what we know

Multimodal measurement in neuroscience has become a mainstream experiment setting since it can investigate a neural system in different spatiotemporal scales to reveal the data generation mechanism across levels. For example, the simultaneous acquisition of functional magnetic resonance imaging (fMRI) and electrophysiology (Ephy) enables us to study the causal link - Hemodynamic Response Function (HRF), from the Ephy Local Field Potential (LFP) signal to the fMRI blood oxygenation level dependent (BOLD) signal.

(2) Introduce the question and what we don’t know

However, the measured signal in the local system is not only contaminated by exogenous independent noise sources, but also influenced by the local module interaction and the global environment state due to the nested structure of neural systems. In the case of concurrent fMRI-Ephy recording, the raw Ephy data is not only mixed with the fMRI gradient interference, but also contains the interaction effect of local frequency bands and the modulation effect of the global brain state. It’s still challenging to separate noise components for the high quality denoised Ephy data and to estimate the local HRF without bias, which results from the high-dimensional confounders inlcuding other local frequency bands and the global fMRI BOLD signal.

(3) provide major methods and results

Following the lab rotation work, we develop an ensemble method to remove fMRI gradient artifacts, which also integrates a new metric to evaluate the denoised signal performance qualitatively and quantitively. The denoised method consists of multiple decomposition algorithms, including Principal Component Analysis (PCA), Independent Component Analysis (ICA) and Non-negative Matrix Factorization (NMF), and the evaluation metric is based on the coherence between recovered spike trains and the LFP during the detected ripple events. While the comparative study shows that the optimized setting in the denoised pipeline are dependent on*** data itself characteristics, we provide empirical initializations and automatic parameter selection for the general purpose.

Afterwards, we build a realistic model of multimodal signals to estimate the local HRF in a nonparametric way, by applying Double Machine Learning (DML) which combines standard causal inference and advanced machine learning (ML) techniques. The DML estimator is theoretical unbiased for estimating the interested treatment effect (TE) from one local LFP band to the local BOLD signal, i.e., the low-dimensional parameters in the local HRF, by removing all nuisance effects of confounders. In this thesis, we assume a linear time-invariant (LTI) system between the neural activity and the BOLD response, and implement the multi-treatment DML via EconML toolbox, where the local HRF is approximated by the LTI system’s impulse response function.

(4) final conclusions and how can we get closer

Our experimental results demonstrate that the ensemble denoised approach can significantly reduce various noise effects to obtain high-quality Ephy signals with increased Signal-to-Noise Ratio (SNR) of ripple events, compared to a standard averaged artifact subtraction (AAS) method. On the simulated and real dataset, we show that the DML method is always a fair choice compared to the naive method for the local HRF estimation, though it’s more computational expensive and the performance improvement is actually determined by the underlying model relationships. Given the complexity and nonlinearity of neural systems, we still recommend the DML method to estimate causal parameters and draw statistical inference about the local HRF, especially when multimodal signals of potential confounders are available.

Summary

(1) Ephy Denoising

In summary, we investigated an ensemble denosing toolkit for the elimination of various noise effects on the Ephy signal during Ephy-fMRI experiment, which is mainly corrupted by the gradient interference noise of the fMRI imaging. We show the detailed denoising workflow in two representative sessions K13m17 and E10bv1, and demonstrate that for clean dataset, (AAS plus) PCA and ICA are generally adequate and for datasets contaminated with large or complex noise structures, NMF would be a better choice. We also suggest some principles of how to inspect and evaluate the denoised signal quality, from the time plots, spectral plots to the spike-LFP coherence of ripple events in the hippocampus.

(2) HRF Estimation

Overall, we build a LTI model with confounders for the causal HRF estimation by using the naive and DML method. DML is able to correct the regularization bias entailed by the ML prediction of high-dimensional confounders. Thus, the DML method is suitable to estimate site-specific HRF, i.e., the treatment effects from the one local frequency band to the local BOLD signal. The simulation results show the effects of multi-treatments with multi-confounders, data normalization, model selection, and parameter tuning. The experimental results show that in the dataset E10aw1, the gamma and ripple bands are the primary treatments and the global fMRI signal is the most effective confounders, compared to the other local frequency bands. Together, it demonstrates that the DML method is a worthy choice to estimate the HRF compared to the naive regression method, especially when there are lots of potential confounders with the nonlinear relationship.

Presentation

Hong’s master thesis

Acknowledgments

I would like to express my thanks to Dr. Michel Besserve. When I first met him in the signal processing lecture, I was impressed by his organized slide and passionate teaching. After the lab visit, I admire him for his marvellous work, which is extensive in the interdisciplinary area among neuroscience, causal analysis and machine learning. I’m really excited to work with him to explore the advanced topics and techniques, including how sleep rhythms affect memory consolidation, how to denoise fMRI interference by nonnegative matrix factorization, and how to estimate the causal interactions by double machine learning. He guides me the research direction by his intelligent mind regularly but also gives me flexibility to consider my interests. I always enjoy the conversation with him, when he points out the key question in the office or discusses the new ideas in the Skype. I learn a lot from him during the time of essay rotation, lab rotation and master thesis, and I’m keen to be a creative and productive researcher in the future.

At the same time, I’m grateful that I have some great colleagues and friends. Kaidi Shao is the one who introduces me the lab work and enriches my life, including delicious food and gossiping. She helps me go through the paper work and shares her valuable experience, like how to handle the PC with a hidden helpful man - Joachim Werner. Shervin Safavi influences me by his professional style, generously providing constructive advice and actively sharing academic news about paper, talk and conference. I will remember all the time with their support and company.

2020 Michel Group Skype

1. Institute of Neurobiology, University of Tübingen, Tübingen 72076, Germany,
2. Werner-Reichardt Centre for Integrative Neuroscience, Tübingen 72076, Germany,
3. Graduate Training Centre of Neuroscience, IMPRS, Tübingen 72074, Germany

Download

screenshot from Jneurosci webpage

Abstract

Plasticity within hippocampal circuits is essential for memory functions. The hippocampal CA2/CA3 region is thought to be able to rapidly store incoming information by plastic modifications of synaptic weights within its recurrent network. High-frequency spike-bursts are believed to be essential for this process, by serving as triggers for synaptic plasticity. Given the diversity of CA2/CA3 pyramidal neurons, it is currently unknown whether and how burst activity, assessed in vivo during natural behavior, relates to principal cell heterogeneity. To explore this issue, we juxtacellularly recorded the activity of single CA2/CA3 neurons from freely-moving male mice, exploring a familiar environment. In line with previous work, we found that spatial and temporal activity patterns of pyramidal neurons correlated with their topographical position. Morphometric analysis revealed that neurons with a higher proportion of distal dendritic length displayed a higher tendency to fire spike-bursts. We propose that the dendritic architecture of pyramidal neurons might determine burst-firing by setting the relative amount of distal excitatory inputs from the entorhinal cortex.

Acknowledgments

Loading wonderful memories in Burgalossi’s lab…

Abstract

The simultaneous acquisition of functional magnetic resonance imaging (fMRI) and electrophysiology (Ephy) data is a promising experiment technique in neuroscience because it allows us to study the same neural system from two different levels concurrently. However, recording fMRI-Ephy together induces fMRI gradient interference on Ephy data and denoising Ephy signals is still challenging. Here we review methods of reducing gradient artifacts, including hardware-based and software-based, and present a new denoising method, Non-negative Matrix Factorization with Itakura-Saito divergence (IS-NMF). Our results demonstrate that IS-NMF approach can effectively remove residual artifacts to obtain high quality Ephy data after applying hardware compensation and standard averaged artifact subtraction (AAS).

Summary

In summary, when we record fMRI and Ephy signals together, the Ephy data contains not only the original neural electrical signals but also fMRI-induced gradient artifacts and residual noise. After using hardware circuits to compensate the fMRI interference, the rest of noise can be eliminated by performing IS-NMF in addition to AAS method to get the denoised Ephy data.

Presentation

Hong’s lab rotation report

Abstract

The major function of sleep, system memory consolidation, is believed to achieve by regulating synaptic strength during sleep-state-dependent oscillations. In recent years, experimental evidence shows the up and down effects of sleep oscillations, supporting and against the Synaptic Homeostasis Hypothesis (SHY) that predicts the sleep effects on synaptic strength. Although in 2017 Levenstein proposes a model to explain slow wave effects on the population firing distribution, the mechanisms of single sleep wave and its temporal coordination effect are still unclear. Here I review experiment evidence and computational models of sleep oscillation effects, ranging from single rhythm like NREM SWA effect to multiple nested rhythms like SWA-Spindle, SWA-Ripple, and SWA-Spindle-Ripple coupling effect. In order to understand the mechanism of sleep effect on system memory consolidation, we need a memory experiment task which can collect training data - neural population state activities (including network structure, i.e., involved neuron population and connections), and training algorithm - synaptic plasticity rules (synaptic strength modification rules, like General Oscillation Resonance model). And we also need to find out the sub-function to function links for three abstraction levels - behavior, neural population activities, and synaptic connection weights.

Summary

Sleep rhythms have a critical and active role in system memory consolidation, especially the oscillation temporal coordination such as the coupling of SWA, spindle, and ripples in NREM sleep. The underlying mechanism of sleep oscillation effects on synaptic strength is related to the synaptic modification rules like STDP and the neural firing resonance due to different sleep rhythm regulations.

Presentation

Hong’s essay rotation report

Download

See the paper (2010) Evidence for grid cells in a human memory network

The authors: Christian F. Doeller, Caswell Barry & Neil Burgess

Summary

In general, the authors transfer electrophysiological experiemnts in rats to virtual reality fMRI recordings in humans to show that humans have grid-cell-like representations in a network of entorhinal areas which support spatial cognition like rats.

Presentation

Hong’s retreat presentation

Computer Science 1303 Persuasive Speech

Speech Title/Topic: Give the right of choosing roommates to students.

General Purpose: To persuade.

Specific Purpose: To persuade the audience that current system of dormitory assignment has great shortages. After hearing my speech, the audience will agree that there is a need of reform and they will take actions.

Intended Audience Outcome: To convince the audience of my claim and urge them to promote the reform.

Organizational Structure: Problem-Cause-Solution Pattern, Claim of Policy

Thesis: The method of dormitory assignment is inconsiderate at many Chinese universities. The system needs to be reformed.

I. Introduction

1. Attention Getter

I would like to begin my speech by asking you some questions here: Have you ever got mad by your roommates’ loud music at midnight? Have you ever been annoyed by your roommates’ alarm clock buzzing in early morning? Have you ever thought why you were there? Have you ever thought what if you could choose your roommates at the beginning?

2. Narration

If you answers are yes, you are not alone. China Youth Daily reports that about 68% interviewed university students have considered asking to change their dormitories because of the roommate conflicts. When they were asked about the best way to match roommates, the first answer is to match roommates based on their own personal lifestyle and their roommate preferences. There is no doubt that the interpersonal relationship in dormitory will significantly influence the quality of campus life which students experience for four years. The tense roommate relationships can harm students’ well-being in both study and life.

3. Thesis&&Credibility

I deeply know how much harm it can do because I was once a victim. When I could not sleep due to my roommate’s loud music, I was thinking about why I was there and try to find a radical solution. After reading a score of research papers for this topic and making a class survey, I am surprised to find that the inconsiderate dormitory assignment is a very common phenomenon and many schools pay scant attention to it for a long time.

4. Preview

Tonight, we are going to discuss the problems caused by the current housing system at our school, and explore some effective solutions. Let us start by looking at the need for the reform of the dormitory assignment rules.

II. Body

1. Main Point A

The current dormitory assignment system has great shortage, which can harm students’ physical and psychological growth.

a. As we all know, every year incoming freshmen at northeastern university will be placed into dormitories mainly based on their majors and classes. If you are lucky, you will be matched with roommates who have compatible lifestyles with yours and you will have a good time in your harmonious dormitory. Unfortunately, using this random assignment method, most of the students will suffer an issue of living with people who have incompatible lifestyles with them, which tends to cause conflicts in dormitory.

• My class survey shows that 83 percent of you have been so annoyed by your roommates that you wanted to live in another room. In this case, the roommate relationship will become tenser and tenser because of the disputes between roommates about life habits such as bed times and noise levels in the room. And a strained dormitory environment can harm students’ physical and psychological well-being.
• According to a paper about the relationship between roommate friendliness and social skills development, a poor roommate relationship can cause more physical illness, more psychological stress and less involvement in college activities.¹

b. What is worse, when you are suffering and intend to apply to change your room, the current dormitory regulation provides you with a complicated procedure and little chance of success. So the students have to endure a painful living circumstance where the accumulated complaints and decreased tolerance among roommates may lead to extreme tragedies.

• It is hard to forget Fudan university poisoning case. In 2013, a Fudan University graduate student, Huang Yang, was poisoned to death by his roommate because of disharmonious roommate relationship.²

Transition: We have seen the seriousness of the problems caused by the current dormitory assignment system, now let us explore some solutions.

2. Main Point B

Try to explore what is a reasonable dormitory assignment system.

a. Let me make a clarification first. Is a quiet person more admirable than a communicative person? Are early risers better than night owls? I do not know and we do not discuss them here. Because everybody has their own version of what is a satisfying life, there is nothing wrong with them. The problem is the random assignment rule, which is likely to assign these incompatible people the same room.

b. So how to design an ideal rule to avoid the problem? Well, I think we can draw on the experience of foreign universities such as Cornell University where students will be matched with roommates with compatible lifestyle preference profiles which they complete in the housing application.³

For decades, some famous universities like Stanford believed that putting people in uncomfortable situations with a complete stranger was an essential learning part of college. But in recent years, universities have moved to more modern systems - by 2012, about 70 percent of schools have allowed incoming freshmen to select roommates based on a preference survey. Julie Weber, director of housing at New Mexico State, said “After choosing roommate by themselves, the number of students who asked to switch roommates dropped 40 percent, and the students’ GPAs were higher.”⁴

c. So from the experience of foreign universities, we can naturally conclude that an ideal dormitory assignment rule must include three parts as follows.

• First, before incoming freshmen arrive on campus, they will be asked to list out personal preferences and roommate preferences in the housing application. Students can request that his/her roommates should not be smokers, although he/she is a smoker. Maybe someone does not care about whether their roommates are smokers.
• Second, after school matches roommates based on the application, students have to live with their roommates for one week called trial period. This section is designed for students to give their room assignment a chance and for schools to have enough time to complete a room census.
• Third, after one week trial period, schools should provide a convenient room change procedure for students who still cannot get along with their roommates, called green channel.

d. In fact, some domestic universities have begun to reform the dormitory assignment system following the above principles. And it also has proved that the reform is practical.

• In 2014, Shanghai Jiao Tong University declared that every incoming freshman will be asked to list their preferences like sleeping habit for the room arrangement.⁵
• In 2015, Nanjing University of Science and Technology also announced that freshmen can use roommate matching program to help select their roommates. These reforms have made a promising results that students are more satisfied with dormitories and have more energy to take on campus life.⁶

III. Conclusion

1. Review

In conclusion, the current dormitory assignment rules should be reformed because it is likely to cause a tense living situation which will harm the students’ physical and psychological health.

2. Restatement of Thesis

An ideal assignment system must be designed to help students choose their roommates, including preference survey, trial period and green channel.

3. Peroration

To achieve the reform, I need your help. The first thing I need you to do is to let more people know about this idea, especially students. It is a truth that everyone has the right to pursue their happiness. Since you can choose your university, your major, your work, why you cannot choose your roommates? The second thing I need you to do is to send a letter to our school leaders. This thing is simple because NEU has a website called President Mail Box, but it need some courage. However, I am glad to see from my class survey that 67 percent of you are brave enough to propose this idea to our headmaster. And I will be the first one to send the letter! Let us take action now!

There will be no complaint in dormitory anymore.

Let the school hear our voice! We will be remembered. Thank you!

References

There is no doubt that the development of the Internet has made our lives more convenient than ever before, but it also has brought some problems.

In 2013, QQ was found that more than 70 million data had leaked on the Internet. In 2014, 12306 website was accused that about 130 thousand personal information had been public. Last month, NetEase 163 mail had a big deal. Over 100 million 163 emails were hacked.

Since my major is computer science and technology and I have been interested in hacking skills for over three years, tonight, I would like to share something related to it. - Internet Security

According to the data breach investigations report, in 2015, 700 million personal records were stolen, sold, and made public by crackers. Crackers are bad hackers. Hackers work for skills and fun, while crackers work for money. They like to steal your network information and ruin it. Because the number of crackers has increased in recent years, the Internet has become more and more dangerous.

So how can we protect our computers from crackers?

(Well) I think the best way to learn how to defend is to learn how to attack. Once you know how crackers steal your data, you can avoid it. So I am going to tell you two common ways to steal other information in our daily life. Let us start password first.

Password

Password, the only identification on the virtual world, from QQ to online bank, we use it everywhere. Obviously, it is very important, but it could be fragile if you are careless. If you want to crack other passwords, what should you do? It’s easier than you think.

1. The first thing you can do is to try some default passwords which are set by system, such as admin, 1234, or 0000. Ridiculously? Yes, but it is very effective. Believe it or not, more than half of the people never change their passwords, because they are so lazy. Lazy, is one key to crack passwords.
2. If you are unlucky, the next step is to try weak password. By contrast with the strong password, weak password is easy to remember and guess. As the picture shows, here are some common weak password. By using the specific software, you can guess the password over Ten million times in one second.
3. Fail again? Do not worry, let us try to crack his/her other accounts first, especially the unimportant accounts. Once you get one account password, you can apply a hacker skill called library crash for obtaining his/her other accounts. Because people are likely to use the same or similar password at different sites.

For example, I suppose your Wechat password is the same as your QQ password. Yeah? Got it? Remember the principle is that people are Lazy. Let us move on to Wi-Fi.

Wi-Fi

Wi-Fi could be dangerous, too. When you think you are lucky to have a free Wi-Fi, maybe someone are spying on your information through the Free Wi-Fi.

• To do that, first, set a free Wi-Fi hotspot named CMCC or NEU.
• And wait people to connect your Wi-Fi happily, something like fishing.
• Then run a analysis tool named sniffer to catch data he send to the server so you can steal his cookies and sign in his accounts. Because most people always open their Wi-Fi which will lead to automatic connection, we can monitor their data and catch their information packet easily.

Conclusion

Now you know two ways to steal information. On the one hand, you can attack passwords through default passwords, weak password and library crash. On the other hand, you can attack Wi-Fi by building a free fishing hotspot. But don’t forget our original purpose, to protect ourselves information. So the contrary ways are to set different strong passwords at different websites and to be wary of Free Wi-Fi.

Keep your eye at your personal information. Thank you!